U.S. patent application number 16/718964 was filed with the patent office on 2022-08-18 for controlling a jitter buffer.
This patent application is currently assigned to DOLBY LABORATORIES LICENSING CORPORATION. The applicant listed for this patent is DOLBY LABORATORIES LICENSING CORPORATION. Invention is credited to Zhiwei SHUANG, Xuejing SUN.
Application Number | 20220263423 16/718964 |
Document ID | / |
Family ID | 1000006505000 |
Filed Date | 2022-08-18 |
United States Patent
Application |
20220263423 |
Kind Code |
A9 |
SUN; Xuejing ; et
al. |
August 18, 2022 |
CONTROLLING A JITTER BUFFER
Abstract
Apparatus and methods for controlling a jitter buffer are
described. In one embodiment, the apparatus for controlling a
jitter buffer includes an inter-talkspurt delay jitter estimator
for estimating an offset value of the delay of a first frame in the
current talkspurt with respect to the delay of a latest anchor
frame in a previous talkspurt, and a jitter buffer controller for
adjusting a length of the jitter buffer based on a long term length
of the jitter buffer for each frame and the offset value.
Inventors: |
SUN; Xuejing; (Beijing,
CN) ; SHUANG; Zhiwei; (Beijing, CN) |
|
Applicant: |
Name |
City |
State |
Country |
Type |
DOLBY LABORATORIES LICENSING CORPORATION |
San Francisco |
CA |
US |
|
|
Assignee: |
DOLBY LABORATORIES LICENSING
CORPORATION
San Francisco
CA
|
Prior
Publication: |
|
Document Identifier |
Publication Date |
|
US 20200127577 A1 |
April 23, 2020 |
|
|
Family ID: |
1000006505000 |
Appl. No.: |
16/718964 |
Filed: |
December 18, 2019 |
Related U.S. Patent Documents
|
|
|
|
|
|
Application
Number |
Filing Date |
Patent Number |
|
|
14654346 |
Jun 19, 2015 |
10560393 |
|
|
PCT/US13/76387 |
Dec 19, 2013 |
|
|
|
16718964 |
|
|
|
|
61759023 |
Jan 31, 2013 |
|
|
|
Current U.S.
Class: |
1/1 |
Current CPC
Class: |
H02M 7/4807 20130101;
H02M 1/42 20130101; H02M 7/79 20130101; H02M 1/14 20130101; B60L
55/00 20190201; Y02B 70/10 20130101; H02M 5/458 20130101; B60L
53/14 20190201; B60L 1/00 20130101 |
International
Class: |
H02M 5/458 20060101
H02M005/458; B60L 1/00 20060101 B60L001/00; H02M 7/48 20060101
H02M007/48; H02M 7/79 20060101 H02M007/79; H02M 1/42 20060101
H02M001/42; H02M 1/14 20060101 H02M001/14; B60L 55/00 20060101
B60L055/00; B60L 53/14 20060101 B60L053/14 |
Foreign Application Data
Date |
Code |
Application Number |
Dec 20, 2012 |
CN |
201210560747.3 |
Claims
1. A method of controlling a jitter buffer, the method comprising:
estimating a long term length of the jitter buffer for each frame
by calculating a statistic distribution of history delay jitter
values; and setting the length of the jitter buffer for the first
frame of a current talkspurt based on the long term length for the
first frame and the number of frames received at the same time with
the first frame.
2. The method of claim 1, further comprising: in response to
determining that the jitter buffer length still used by the last
talkspurt is greater than the long term length and the first frame
of a new talkspurt has arrived, dropping one or more frames of the
last talkspurt so that the frames of the last talkspurt occupy only
the long term length of the jitter buffer.
3. The method of claim 1, further comprising adjusting the present
length of the jitter buffer toward the long term length.
4. The method of claim 3, wherein adjusting the present length of
the jitter buffer toward the long term length comprises calculating
a weighted average of the long term length and the present
length.
5. The method of claim 4, wherein the weight of the long term
length is greater when the long term length is greater than the
present length, and is smaller when the long term length is smaller
than the present length.
6. A system for controlling a jitter buffer, the system comprising:
one or more processors; and a non-transitory computer-readable
medium storing instructions that, upon execution by the one or more
processors, cause the one or more processors to perform operations
comprising: estimating a long term length of the jitter buffer for
each frame by calculating a statistic distribution of history delay
jitter values; and setting the length of the jitter buffer for the
first frame of a current talkspurt based on the long term length
for the first frame and the number of frames received at the same
time with the first frame.
7. The system of claim 6, the operations further comprising: in
response to determining that the jitter buffer length still used by
the last talkspurt is greater than the long term length and the
first frame of a new talkspurt has arrived, dropping one or more
frames of the last talkspurt so that the frames of the last
talkspurt occupy only the long term length of the jitter
buffer.
8. The system of claim 6, the operations further comprising
adjusting the present length of the jitter buffer toward the long
term length.
9. The system of claim 8, wherein adjusting the present length of
the jitter buffer toward the long term length comprises calculating
a weighted average of the long term length and the present
length.
10. The system of claim 9, wherein the weight of the long term
length is greater when the long term length is greater than the
present length, and is smaller when the long term length is smaller
than the present length.
11. A non-transitory computer-readable medium storing instructions
that, upon execution by one or more processors, cause the one or
more processors to perform operations of controlling a jitter
buffer, the operations comprising: estimating a long term length of
the jitter buffer for each frame by calculating a statistic
distribution of history delay jitter values; and setting the length
of the jitter buffer for the first frame of a current talkspurt
based on the long term length for the first frame and the number of
frames received at the same time with the first frame.
12. The non-transitory computer-readable medium of claim 11, the
operations further comprising: in response to determining that the
jitter buffer length still used by the last talkspurt is greater
than the long term length and the first frame of a new talkspurt
has arrived, dropping one or more frames of the last talkspurt so
that the frames of the last talkspurt occupy only the long term
length of the jitter buffer.
13. The non-transitory computer-readable medium of claim 11, the
operations further comprising adjusting the present length of the
jitter buffer toward the long term length.
14. The non-transitory computer-readable medium of claim 13,
wherein adjusting the present length of the jitter buffer toward
the long term length comprises calculating a weighted average of
the long term length and the present length.
15. The non-transitory computer-readable medium of claim 14,
wherein the weight of the long term length is greater when the long
term length is greater than the present length, and is smaller when
the long term length is smaller than the present length.
Description
CROSS-REFERENCE TO RELATED APPLICATIONS
[0001] This application is a divisional of U.S. patent application
Ser. No. 14/654,346, filed on 19 Jun. 2015, which is a National
Stage Entry of PCT/US2013/076387 filed on 19 Dec. 2013, which
claims priority to Chinese Patent Application No. 201210560747.3
filed 20 Dec. 2012 and U.S. Provisional Patent Application No.
61/759,023 filed 31 Jan. 2013, which is hereby incorporated by
reference in its entirety.
TECHNICAL FIELD
[0002] The present application relates generally to audio signal
processing. More specifically, embodiments of the present
application relate to apparatus and methods for controlling a
jitter buffer.
BACKGROUND
[0003] Voice transmission over packet networks is subject to delay
variation, commonly known as jitter, due to highly heterogeneous
network conditions. In IP-based networks, fixed delay can be
attributed to algorithmic, processing and propagation delays due to
material and distance, whereas variable delay is caused by the
fluctuation of IP network traffic, different transmission path over
the internet, etc. VoIP (voice over Internet Protocol) receivers
generally rely on a "jitter buffer" to counter the negative impact
of jitter. Here the packet delay variation, or jitter, is defined
as the difference between the actual arrival time of the packets
and a reference clock at the normal packet rate. By introducing an
additional "playout" delay, a jitter buffer aims at transforming
the uneven flow of arriving packets into a regular flow of packets,
such that delay variations will not cause perceptual quality
degradation to the end users. Voice communication is highly delay
sensitive, where one-way delay should be kept below 150 ms for
normal conversation, with above 400 ms being considered
unacceptable (ITU Recommendation G114). Therefore, the additional
delay added by a jitter buffer needs to be as small as possible.
Unfortunately, a small jitter buffer will lead to more frequent
packet loss when packets arrive later than their expected playout
deadline due to jittered network.
[0004] Various jitter buffer management methods have been proposed
in an aim to find an optimal balance between delay and packet loss.
In a typical jitter buffer algorithm, the incoming packets are
stored in a buffer, along with meta-data, like sequence numbers
etc. After certain amount of time, e.g. 20 ms, the oldest packet
from the jitter buffer will be sent to the decoder. New packets are
inserted into the jitter buffer based on their sequence number
accordingly. Two general approaches exist in jitter buffer design,
namely fixed jitter buffer and adaptive jitter buffer. A jitter
buffer with fixed size provides a constant buffer delay and incurs
minimum complexity. Obviously it can introduce excessive delay if
large delay variations need to be accommodated. Adaptive jitter
buffer allocates the size dynamically, which are more commonly used
in practice.
[0005] There is a need to further improve the jitter buffer
management algorithm such that an optimal overall perceived quality
can be achieved.
SUMMARY
[0006] According to an embodiment of the application, an apparatus
for controlling a jitter buffer is provided, including an
inter-talkspurt delay jitter estimator for estimating an offset
value of the delay of a first frame in the current talkspurt with
respect to the delay of a latest anchor frame in a previous
talkspurt and a jitter buffer controller for adjusting a length of
the jitter buffer based on a long term length of the jitter buffer
for each frame and the offset value.
[0007] According to another embodiment of the application, an
apparatus for controlling a jitter buffer includes a long term
length estimator for estimating a long term length of the jitter
buffer for each frame by calculating a statistic distribution of
history delay jitter values; and a jitter buffer controller for
setting the length of the jitter buffer. In the apparatus, the
jitter buffer controller is configured to set the length of the
jitter buffer for the first frame of the current talkspurt based on
the long term length for the first frame and the number of frames
received at the same time with the first frame.
[0008] Another embodiment of the application involves an apparatus
for controlling a jitter buffer, which includes a long term length
estimator for estimating a long term length of the jitter buffer
for each frame by calculating a statistic distribution of history
delay jitter values; and a jitter buffer controller for adjusting
the present length of the jitter buffer toward the long term
length.
[0009] Yet another embodiment of the application provides an
apparatus for controlling a jitter buffer, including a jitter
buffer monitor for monitoring occupancy of the jitter buffer, and a
jitter buffer controller for increasing the length of the jitter
buffer in response to the occupancy meeting a predefined
condition.
[0010] The application also provides a method for controlling a
jitter buffer. According to one embodiment of the method, it
includes estimating an offset value of the delay of a first frame
in the current talkspurt with respect to the delay of a latest
anchor frame in a previous talkspurt; and adjusting a length of the
jitter buffer based on a long term length of the jitter buffer for
each frame and the offset value.
[0011] According to another embodiment of the method, a long term
length of the jitter buffer for each frame is estimated by
calculating a statistic distribution of history delay jitter
values; and the length of the jitter buffer for the first frame of
the current talkspurt is set based on the long term length for the
first frame and the number of frames received at the same time with
the first frame.
[0012] Another embodiment of the method involves estimating a long
term length of the jitter buffer for each frame by calculating a
statistic distribution of history delay jitter values; and then
adjusting the present length of the jitter buffer toward the long
term length.
[0013] Yet another embodiment of the method includes monitoring
occupancy of the jitter buffer, and then increasing the length of
the jitter buffer in response to the occupancy meeting a predefined
condition.
BRIEF DESCRIPTION OF DRAWINGS
[0014] The present application is illustrated by way of example,
and not by way of limitation, in the figures of the accompanying
drawings and in which like reference numerals refer to similar
elements and in which:
[0015] FIG. 1A is a diagram schematically illustrating an exemplary
voice communication system where embodiments of the application can
be applied;
[0016] FIG. 1B is a diagram schematically illustrating another
exemplary voice communication system where embodiments of the
application can be applied;
[0017] FIG. 2 is a diagram illustrating an apparatus for
controlling a jitter buffer according to an embodiment of the
application;
[0018] FIG. 3 is a diagram illustrating selection of an anchor
frame and calculation of inter-talkspurt delay jitter;
[0019] FIG. 4 is a diagram illustrating an apparatus for
controlling a jitter buffer according to a variation of the
embodiment as shown in FIG. 2;
[0020] FIG. 5 is a diagram illustrating an apparatus for
controlling a jitter buffer according to another variation of the
embodiment as shown in FIG. 2;
[0021] FIG. 6 is a diagram illustrating an apparatus for
controlling a jitter buffer according to another variation of the
embodiment as shown in FIG. 2;
[0022] FIG. 7 is a diagram illustrating an apparatus for
controlling a jitter buffer according to yet another variation of
the embodiment as shown in FIG. 2;
[0023] FIG. 8 is a diagram illustrating an apparatus for
controlling a jitter buffer according to another embodiment of the
application;
[0024] FIG. 9 is a diagram illustrating an apparatus for
controlling a jitter buffer according to a variation of the
embodiment as shown in FIG. 8;
[0025] FIG. 10 is a diagram illustrating another embodiment of the
application;
[0026] FIG. 11 is a diagram illustrating an apparatus for
controlling a jitter buffer according to yet another embodiment of
the application;
[0027] FIG. 12 is a diagram illustrating an apparatus for
controlling a jitter buffer according to a variation of the
embodiment as shown in FIG. 11;
[0028] FIG. 13 is a block diagram illustrating an exemplary system
for implementing embodiments of the present application;
[0029] FIG. 14 is a flow chart illustrating setting and updating of
anchor frames according to embodiments of the method for
controlling a jitter buffer of the present application;
[0030] FIG. 15 is a flow chart illustrating estimating of the long
term length of the jitter buffer according to embodiments of the
method for controlling a jitter buffer of the present
application;
[0031] FIG. 16 is a flow chart illustrating estimating of
inter-talkspurt offset value and adjusting of the long term length
of the jitter buffer according to embodiments of the method for
controlling a jitter buffer of the present application;
[0032] FIG. 17 is a flow chart illustrating updating of jitter
buffer length according to embodiments of the method for
controlling a jitter buffer of the present application; and
[0033] FIG. 18 is a flow chart illustrating pre-emptive expansion
of the jitter buffer according to embodiments of the method for
controlling a jitter buffer of the present application.
DETAILED DESCRIPTION
[0034] The embodiments of the present application are below
described by referring to the drawings. It is to be noted that, for
purpose of clarity, representations and descriptions about those
components and processes known by those skilled in the art but not
necessary to understand the present application are omitted in the
drawings and the description.
[0035] As will be appreciated by one skilled in the art, aspects of
the present application may be embodied as a system, a device
(e.g., a cellular telephone, a portable media player, a personal
computer, a server, a television set-top box, or a digital video
recorder, or any other media player), a method or a computer
program product. Accordingly, aspects of the present application
may take the form of an hardware embodiment, an software embodiment
(including firmware, resident software, microcodes, etc.) or an
embodiment combining both software and hardware aspects that may
all generally be referred to herein as a "circuit," "module" or
"system." Furthermore, aspects of the present application may take
the form of a computer program product embodied in one or more
computer readable mediums having computer readable program code
embodied thereon.
[0036] Any combination of one or more computer readable mediums may
be utilized. The computer readable medium may be a computer
readable signal medium or a computer readable storage medium. A
computer readable storage medium may be, for example, but not
limited to, an electronic, magnetic, optical, electromagnetic,
infrared, or semiconductor system, apparatus, or device, or any
suitable combination of the foregoing. More specific examples (a
non-exhaustive list) of the computer readable storage medium would
include the following: an electrical connection having one or more
wires, a portable computer diskette, a hard disk, a random access
memory (RAM), a read-only memory (ROM), an erasable programmable
read-only memory (EPROM or Flash memory), an optical fiber, a
portable compact disc read-only memory (CD-ROM), an optical storage
device, a magnetic storage device, or any suitable combination of
the foregoing. In the context of this document, a computer readable
storage medium may be any tangible medium that can contain, or
store a program for use by or in connection with an instruction
execution system, apparatus, or device.
[0037] A computer readable signal medium may include a propagated
data signal with computer readable program code embodied therein,
for example, in baseband or as part of a carrier wave. Such a
propagated signal may take any of a variety of forms, including,
but not limited to, electro-magnetic or optical signal, or any
suitable combination thereof.
[0038] A computer readable signal medium may be any computer
readable medium that is not a computer readable storage medium and
that can communicate, propagate, or transport a program for use by
or in connection with an instruction execution system, apparatus,
or device.
[0039] Program code embodied on a computer readable medium may be
transmitted using any appropriate medium, including but not limited
to wireless, wired line, optical fiber cable, RF, etc., or any
suitable combination of the foregoing.
[0040] Computer program code for carrying out operations for
aspects of the present application may be written in any
combination of one or more programming languages, including an
object oriented programming language such as Java, Smalltalk, C++
or the like and conventional procedural programming languages, such
as the "C" programming language or similar programming languages.
The program code may execute entirely on the user's computer as a
stand-alone software package, or partly on the user's computer and
partly on a remote computer or entirely on the remote computer or
server. In the latter scenario, the remote computer may be
connected to the user's computer through any type of network,
including a local area network (LAN) or a wide area network (WAN),
or the connection may be made to an external computer (for example,
through the Internet using an Internet Service Provider).
[0041] Aspects of the present application are described below with
reference to flowchart illustrations and/or block diagrams of
methods, apparatus (systems) and computer program products
according to embodiments of the application. It will be understood
that each block of the flowchart illustrations and/or block
diagrams, and combinations of blocks in the flowchart illustrations
and/or block diagrams, can be implemented by computer program
instructions. These computer program instructions may be provided
to a processor of a general purpose computer, special purpose
computer, or other programmable data processing apparatus to
produce a machine, such that the instructions, which execute via
the processor of the computer or other programmable data processing
apparatus, create means for implementing the functions/acts
specified in the flowchart and/or block diagram block or
blocks.
[0042] These computer program instructions may also be stored in a
computer readable medium that can direct a computer, other
programmable data processing apparatus, or other devices to
function in a particular manner, such that the instructions stored
in the computer readable medium produce an article of manufacture
including instructions which implement the function/act specified
in the flowchart and/or block diagram block or blocks.
[0043] The computer program instructions may also be loaded onto a
computer, other programmable data processing apparatus, or other
devices to cause a series of operational operations to be performed
on the computer, other programmable apparatus or other devices to
produce a computer implemented process such that the instructions
which execute on the computer or other programmable apparatus
provide processes for implementing the functions/acts specified in
the flowchart and/or block diagram block or blocks.
System Overview
[0044] FIG. 1A is a diagram schematically illustrating an example
voice communication system where embodiments of the application can
be applied.
[0045] As illustrated in FIG. 1A, user A operates a communication
terminal A, and user B operates a communication terminal B. In a
voice communication session, user A and user B talk to each other
through their communication terminals A and B. The communication
terminals A and B are coupled through a data link 103. The data
link 103 may be implemented as a point-to-point connection or a
communication network. At either side of user A and user B, VAD
(Voice Activity Detection) is performed on audio blocks of the
audio signal captured by the user's communication terminal. If
voice presence is decided in an audio block, corresponding
processing (e.g., applying a gain suitable for voice) is performed
on the audio block and the audio block is transmitted to another
user's communication terminal A through the data link 103. If no
voice presence is decided in an audio block, corresponding
processing (e.g., applying a gain suitable for non-voice) is
performed on the audio block and the audio block is transmitted to
another user's communication terminal A through the data link 103.
In this case, it is also possible to transmit simple information
indicating a silence period to the other user's communication
terminal, or to indicate that the time period corresponding to this
audio block is silent by transmitting nothing. The other user's
communication terminal receives the transmitted audio blocks and
stores them to entries corresponding to the same time as the audio
blocks in its jitter buffer, to eliminate transmission jitters.
Audio blocks in the jitter buffer are fed through decoding and
processing to reproduce them at the output transducer or
transducers of the communication terminal. Reception of the simple
information or nothing may cause corresponding empty entries in the
jitter buffer.
[0046] FIG. 1B is a diagram schematically illustrating another
example voice communication system where embodiments of the
application can be applied. In this example, a voice conference may
be conducted among users.
[0047] As illustrated in FIG. 1B, user A operates a communication
terminal A, user B operates a communication terminal B, and user C
operates a communication terminal C. In a voice conference session,
user A, user B, and user C talk to each other through their
communication terminals A, B, and C, respectively. The
communication terminals illustrated in FIG. 1B have the same
function as those illustrated in FIG. 1A. However, the
communication terminals A, B, and C are coupled to a server through
a common data link 113 or separate data links 113. The data link
113 may be implemented as a point-to-point connection or a
communication network. At either side of user A, user B, and user
C, VAD is performed on audio blocks of the audio signal captured by
the user's communication terminal. If voice presence is decided in
an audio block, corresponding processing (e.g., applying a gain
suitable for voice) is performed on the audio block and the audio
block is transmitted to the server through the data link 113. If no
voice presence is decided in an audio block, corresponding
processing (e.g., applying a gain suitable for non-voice) is
performed on the audio block and the audio block is transmitted to
the server through the data link 113. In this case, it is also
possible to transmit simple information indicating a silence period
to the server 115, or to indicate that the time period
corresponding to this audio block is silent by transmitting
nothing. The server receives the transmitted audio blocks and
stores them to entries corresponding to the same time as the audio
blocks in its jitter buffers associated with the users
respectively, to eliminate transmission jitters. Audio blocks
corresponding to the same time in the jitter buffers are mixed into
one audio block and the mixed audio block is transmitted to all
users A, B, and C. Reception of the simple information or nothing
may cause corresponding empty entries in the jitter buffers. The
users' communication terminals receive the audio blocks from the
server and store them to entries corresponding to the same time as
the audio blocks in their jitter buffers, to eliminate transmission
jitters. In each communication terminal, audio blocks in the jitter
buffer are fed to a voice processor to reproduce them through the
speaker system of the communication terminal. Although three
communication terminals are illustrated in FIG. 1B, there can
reasonably be two or more communication terminals coupled in the
system.
Estimation of Inter-Talkspurt Jitter (Offset)
[0048] A first embodiment of the present application provides an
apparatus 200 for controlling a jitter buffer 206 as shown in FIG.
2. The apparatus comprises an inter-talkspurt delay jitter
estimator 202 for estimating an offset value of the delay of a
first frame in the current talkspurt with respect to the delay of a
latest anchor frame in a previous talkspurt, and a jitter buffer
controller 204 for adjusting a length of the jitter buffer 206
based on a long term length of the jitter buffer 206 for each frame
and the offset value.
[0049] According to the embodiment, the length (or size, or level)
of the jitter buffer 206 may be adjusted based on the long term
length of the jitter buffer (JB) 206 and the offset value (or
jitter value) of the delay of the first frame in the current
talkspurt. The long term length of the JB reflects a long term
level of delay jitter in history audio signals, and the offset
value of the delay of the first frame reflects the possible delay
jitter of the current talkspurt. Thus the apparatus for controlling
the JB according to the embodiment adjusts the length of the JB 206
based on both.
[0050] Please note that here the term "frame" is used. In the
context of the present application, we assume "jitter buffer" is a
"logical" jitter buffer storing audio frames. While depending on
specific implementations, the physical jitter buffer may store any
forms of packets or audio frames. Therefore, throughout the
specification, the term "jitter buffer" shall be construed as
including both jitter buffer actually storing audio frames and
jitter buffer actually storing any forms of packets (blocks) which
will be decoded into audio frames before being played out or being
fed into any components where audio frames are necessary, and the
decoding process will not be explicitly discussed in the present
application although it does exist. Accordingly, the term "frame"
shall be construed as including a real frame already decoded from a
packet or still encoded in the packet, or a packet itself including
one or more frames, or more than one frame encoded in a packet or
already decoded from the packet. In other words, in the context of
the present application, a processing involving a frame may also be
construed as a processing involving a packet, or a processing
involving simultaneously more than one frame contained in a
packet.
[0051] The long term length of the JB is conventionally available
for each frame. It may be estimated with a long term length
estimator 802 (see FIG. 8) by computing a histogram of the past
jitter values. Alternatively, the histogram can be replaced by a
probability mass function (PMF) (see US patent application
published as US20090003369A1, the contents of which are
incorporated by reference in its entirety). Once the histogram or
PMF is calculated, the long term length can be estimated by setting
a threshold such that the cumulative probability of the expected
delay variation is less than this threshold.
[0052] The offset value of the delay of the first frame is
calculated with respect to the latest anchor frame in the last
talkspurt. Anchor frame is conventionally used as a reference for
estimating a delay jitter of a newly received frame within the same
talkspurt. In the present application, by using the latest anchor
frame in the last talkspurt, a cross-talkspurt or an
inter-talkspurt delay jitter (offset value) is estimated. And thus
the length of the JB may be adjusted upon the start of a new
talkspurt by referring to the last talkspurt.
[0053] The anchor frame in a talkspurt may be set with an anchor
frame setting unit 402 (FIG. 4). Theoretically it is reasonable to
use a frame with minimum one way delay as the anchor frame.
However, it is difficult to measure an absolute one way delay,
requiring both sender and receiver timing information and clock
synchronization. Furthermore, storing many packets in order to get
minimum delay is not practical in a real time system like VoIP, and
using too old an anchor frame may be subject to numerical issues
and clock drift. Therefore, it is proposed to use the first frame
of the talkspurt as the anchor and then keep updating it, so that a
"local" minimum delay within one talkspurt may be used. Another
problem arising from setting the first frame as the anchor frame is
that the first frame might be in a delay spike, which would not be
detected if using the first frame as the anchor. This could lead to
inaccurate delay distribution estimation and over-estimation of
jitter buffer length. Keeping updating the anchor frame within the
talkspurt mentioned above may partly mitigate the problem, and the
problem may be further mitigated through the cross-talkspurt use of
the latest anchor frame in the last talkspurt or more latest anchor
frames in several previous talkspurts as will be discussed later.
In this way, a local minimum delay within a longer time period may
be found and used.
[0054] Therefore, according to a variation 400 of the first
embodiment, the apparatus for controlling a jitter buffer may
further comprise an anchor frame setting unit 402 for initially
setting the first frame in the current talkspurt as the anchor
frame, and setting a newly received frame as the anchor frame when
a delay jitter of the newly received frame meets a predefined
condition. The cross-talkspurt use of the latest anchor frame in
the last talkspurt will be discussed later.
[0055] That is, the anchor frame may be updated in time and thus a
timely-closer anchor frame may be used. For example, if the delay
jitter of a new frame with respect to the present anchor frame is
less than 0, that is, the new frame arrived earlier than expected,
then the new frame may be used as new anchor frame. Or, for
avoiding using too old an anchor frame, the new frame with zero
delay jitter may also be used as the new anchor frame.
[0056] For example, as shown in FIG. 3, there are two talkspurts 1
and 2 separated by a silence frame (shown as a block shadowed with
slash lines). The frames X, Y, Z in Talkspurt 1 are sent at time
t1, t2, t3 (FIG. 3(a)) and would be received on the receiver side
at time t3, t4, t5 with an overall delay (total delay with respect
to the sending time) of two frame gaps (t1 and t2), assuming is a
frame gap of 20 ms and i is an integer (FIG. 3(b)). In FIG. 3(b),
the first frame X would be set as the initial anchor frame. Because
both frame Y and frame Z arrive in time as expected, the delay
jitter thereof with respect to the anchor frame X is zero, X may be
maintained as the anchor frame, or either frame Y or Z may be
updated as new anchor frame (not shown in FIG. 3(b)).
[0057] In the situation shown in FIG. 3(c), frames X and Y arrive
as expected based on the total delay but frame Z arrive 1 frame
(frame gap) earlier than expected, taking frame X or Y as reference
(anchor frame). Thus frame Z is updated as the anchor frame since
the jitter value is less than 0 (-1 frame). Please note that in
FIG. 3(c) and FIG. 3(d), frames Y and Z are shown as arriving at
the same time (t4 or t5, respectively). In the context of the
present application, the meaning of the expression "at the same
time" includes but is not limited to the exact literal meaning, and
shall be construed as "within the same time gap/interval of a
predefined granularity". In the present application, for example,
the predefined granularity may be the time gap between two
consecutively-sent frames/packets (such time gap may be referred to
as frame gap), or network probing rate for checking packet
arrivals, or processing time granularity, but is not limited
thereto. For the latter, e.g. we may quantize the arrival time by
frame duration/gap, e.g. 20 ms. That is, we represent time in
integer number of packets. Similarly, in the context of the present
application, when involving a specific time point, depending on the
context it may also mean a time gap of the predefined granularity.
Further, when involving a specific time t.sub.i (i is integer) for
a certain frame where it shall be understood as a time point,
assuming for clarity that it indicates the time point when the
frame starts on the sender side, or indicates the time point when
the reception of the frame starts on the receiver side.
[0058] In the situation shown in FIG. 3(d), the first frame X is
initially set as anchor frame, with respect to which both frames Y
and Z are further delayed. The delay jitter of frame Y is 2 frames
and that of frame Z is 1 frame, and thus frame X is maintained as
anchor frame until talkspurt 1 completes because the jitter value
is greater than 0.
[0059] Within the current talkspurt, the jitter value of a newly
received frame with respect to the anchor frame may be estimated by
an intra-talkspurt delay jitter estimator 502 (in variation 500 as
shown in FIG. 5). The intra-talkspurt delay jitter estimator 502 is
configured to calculate, as the delay jitter, a difference between
an actual reception time of the newly received frame and an
expected reception time of the same newly received frame, the
expected reception time being calculated with the present anchor
frame as a reference.
[0060] For example, still referring to FIG. 3(c) and assuming the
receiving time of the first frame X (the initial anchor frame) in
talkspurt 1 is t3=R(0), and the expected frame gap is G=20 ms, then
the "expected" receiving time of the following frames Y and Z is
R(0)+20 ms and R(0)+40 ms. While in fact both frames Y and Z are
received at t4=R(0)+20 mses. Thus delay jitter is the difference
between the actual time of receiving respective frames and the
expected time of receiving the respective frames. That is, for
frame Y, the delay jitter is R(0)+20-(R(0)+20)=0; and for frame Z,
the delay jitter is R(0)+20-(R(0)+40)=-20 ms, meaning frame Z
arrived 20 ms earlier. Here, the delay jitter is measured in unit
of time. Since the time gap (frame gap) between two consecutive
frames without delay jitter is fixed (such as 20 ms), the delay
jitter (as well as delay) may also be measured in unit of frame.
For example, we can say the delay jitter of frame Z is -1 frame,
that is, frame Z arrives one frame earlier.
[0061] Within each talkspurt, the anchor frame will be constantly
updated if the predefined condition is met. When each talkspurt
finishes, there is a latest anchor frame, such as frame Z in FIG.
3(c) or frame X in FIG. 3(d) when talkspurt 1 finishes; and for the
new talkspurt, there will be a new anchor frame, such as the first
frame A in FIG. 3 when talkspurt 2 starts. In the present
application, the offset value (jitter) of the delay of the first
frame is calculated with respect to the latest anchor frame in the
last talkspurt, thus the length of the JB may be adjusted upon the
start of a new talkspurt by referring to the last talkspurt.
[0062] In a variation 600 as shown in FIG. 6, the offset value of
the delay of the first frame may be estimated in the
inter-talkspurt delay jitter estimator 202 by means of a time
difference estimator 2022 for calculating a time difference between
the reception time of the latest anchor frame and the reception
time of the first frame, and a frame number estimator 2024 for
estimating an expected number of frames between the latest anchor
frame and the first frame. Specifically, the offset value may be
calculated based on the time difference and an expected time
difference obtained based on the expected number of frames.
[0063] In the frame number estimator 2024, the expected number of
frames may be determined based on sequence numbers of the latest
anchor frame and the first frame and information regarding silence
frames between the previous talkspurt and the current talkspurt,
the information regarding silence frames being carried in at least
one frame in at least one talkspurt in the previous talkspurt and
the current talkspurt.
[0064] For example, in the DTX (Discontinuous Transmission) mode,
no frames are transmitted during silence. The number of non-silence
frames between two non-silence frames may be deduced from the
sequence numbers of the two non-silence frames because non-silence
frames are sequentially numbered regardless whether the two
non-silence frames belong to the same talkspurt. For example, in
FIG. 3, from frame X to frame C, the number of non-silence frames
is C-X=6, meaning C is the fifth frame after frame X (here using X
and C to represent the numbers of the two frames). However, the
number of silence frames lapsed since the end of the last talkspurt
cannot be deduced directly from RTP (Real-time Transport Protocol)
sequence number of RTP packets/frames, because the RTP packet
sequence number will not increment for silence packets/frames. For
example, in FIG. 3, the sequence number of frame A directly follows
the sequence number of frame Z, thus it is impossible to deduce
from the sequence numbers of frames Z and A the number of silence
frames between frames Z and A. Here, DTX mode and RTP format are
just examples and the present application is not limited
thereto.
[0065] Therefore, to calculate the expected number of frames
between two frames in two talkspurts, the number of silence frames
between the two talkspurts must be obtained. For indicating the
information about silence frames, various means may be adopted. For
example, timestamp information may be embedded in all the frames as
that in the standard RTP frame format, or only embedded in the last
frame of the previous talkspurt and the first frame of the current
talkspurt. For example, in FIG. 3, timestamp information may be
embedded in all frames, or only in frames Z and A. In either
situation, the timestamp information in frames Z and A is enough
for deducing the number of silence frames. Specifically, the
timestamp embedded in frame Z will be its sending time t3, and the
timestamp embedded in frame A will be its sending time t5. Then,
the time interval between frames Z and A on the sending side will
be t5-t3-20 ms=20 ms (assuming a frame gap is 20 ms), and it may be
deduced that the number Nsil of silence frames is Nsil=20 ms/20
ms=1 frame. Another alternative is that the number of silence
frames may be embedded in the first frame of the current talkspurt.
For example, in FIG. 3, information indicating that there is one
silence frame may be embedded in frame A, that is, Nsil=1
frame.
[0066] Therefore, in the present application, the information
regarding silence frames may comprise the number of the silence
frames, the number being embedded in the first frame of the current
talkspurt. Or, the information regarding silence frames may
comprise timestamps in the last frame of the previous talkspurt and
the first frame in the current talkspurt.
[0067] Let Nsil denote the number of silence frames between the two
adjacent talkspurts (in the example of FIG. 3, Nsil=1). Then for
the first frame of the current talkspurt, the total expected number
of lapsed frames since the last anchor frame is
Ntotal=S.sub.j(0)-S.sub.j-1(a)+Nsil, where S.sub.j(0) is the
sequence number of the first frame of the current talkspurt j (such
as frame A in talkspurt 2 in FIG. 3), and S.sub.j-1(a) is the
sequence number of the anchor frame of the previous talkspurt j-1
(such as frame Z in talkspurt 1 in FIG. 3(c) or frame X in
talkspurt 1 in FIG. 3(d)). For example, in FIG. 3(c),
Ntotal=A-Z+Nsil=1+1=2 frames, meaning A is the second frame after
the anchor frame Z if taking the silence frame(s) into account. In
FIG. 3(d), Ntotal=A-X+Nsil=3+1=4 frames, meaning A is the fourth
frame after the anchor frame X if taking the silence frame(s) into
account.
[0068] Let G (such as 20 ms) be the expected time gap (frame gap)
between consecutive frames without jitter, then the expected
reception time difference between the latest anchor frame of the
last talkspurt and the first frame of the current talkspurt will be
Ntotal*G. Then the delay jitter of the first frame in the current
talkspurt with respect to the latest anchor frame in the last
talkspurt, that is the inter-talkspurt delay jitter, will be
D.sub.j(0)=R.sub.j(0)-R.sub.j-1(a)-Ntotal*G, where R.sub.j(0) is
the receiving time of the first frame in the current talkspurt j
and R.sub.j-1(a) is the receiving time of the latest anchor frame
of the last talkspurt j-1. In FIG. 3, assuming each t.sub.i (i is
an integer) represents a time gap of 20 ms, then in FIG. 3(c)
D.sub.j(0)=R.sub.j(0)-R.sub.j-1(a)-Ntotal*G=t8-t4-2*20=80-40=40 ms;
and in FIG. 3(d)
D.sub.j(0)=R.sub.j(0)-R.sub.j-1(a)-Ntotal*G=t8-t2-4*20=120-80=40
ms.
[0069] The delay jitter so calculated is measured in unit of time.
For purpose of controlling the size (length) of the jitter buffer,
which is measured in unit of frame, the delay jitter D.sub.j(0) may
be converted into d.sub.j(0) measured in unit of frame by a
division with the time gap (frame gap) G: d.sub.j(0)=.left
brkt-bot.D.sub.j(0)/G.right brkt-bot., where the operation .left
brkt-bot..chi..right brkt-bot. denotes the function of mapping a
real number x to the largest previous integer, that is,
floor(x)=.left brkt-bot..chi..right brkt-bot. is the largest
integer not greater than x. But the operation may also be rounding
or ceiling (ceiling(x)=.left brkt-top..chi..right brkt-bot. is the
smallest integer not less than x). For example, in both FIG. 3(c)
and FIG. 3(d) the delay jitter is d.sub.j(0)=40/20=2 frames. The
following table shows the calculating process discussed above for
FIG. 3(c) and FIG. 3(d), as well as FIG. 3(b):
TABLE-US-00001 Physical meaning of symbols FIG.3 (b) FIG.3 (c)
FIG.3 (d) R.sub.j (0) reception time of 1st frame A in t7 t8 t8
talkspurt 2 R.sub.j-i (a) reception time of the latest t5 (Z) t4
(Z) t2 (X) anchor frame in talkspurt 1 Dr =R.sub.j (0)-R.sub.j-1
(a), real reception t7 - t5 = 40 ms t8 - t4 = 80 ms t8 - t2 = 120
ms time difference S.sub.j (0) sequence number of 1st frame A A A A
in talkspurt 2 S.sub.j-i (a) sequence number of the latest Z Z X
anchor frame in talkspurt 1 S.sub.j (0)-S.sub.j-i (a) number of
non-silence frames 1 1 3 Nsil number of silence frames 1 1 1 Ntotal
=S.sub.j (0)-S.sub.j-1 (a)+Nsil, total 2 2 4 number of frames
Ntotal * G expected reception time 2 * 20 = 40 ms 2 * 20 = 40 ms 2
* 40 = 80 ms difference D.sub.j (0) delay jitter in unit of time 0
40 ms 40 ms d.sub.j (0) =D.sub.j (0)/G, delay jitter in unit of 0 2
2 frame
[0070] Please note that in the example shown in FIG. 3, for clarity
of description, all the delays and jitters are shown as times gaps
which are integer multiple times of the normal frame gap, thus the
calculated delay jitters D.sub.j(0) in unit of time are just
integer multiple times of the frame gap (such as 20 ms). But the
present application is not limited thereto, and the real time
difference between any two frames may not be integer multiple times
of the normal frame gap. In addition, it may be noticed that in
FIG. 3, for facilitating understanding, absolute delay values are
marked; but just as shown in the calculating process discussed
above, such values are unnecessary for the present invention and we
just care the relative delay (that is jitter) values with respect
to the anchor frame.
[0071] In the first embodiment and its variations discussed above,
when estimating the inter-talkspurt delay jitter, that is the
offset value of the delay of the first frame of the current
talkspurt, the latest anchor frame in the last talkspurt is used.
The last talkspurt is a previous talkspurt immediately previous to
the current talkspurt.
[0072] Alternatively, the offset value may be estimated with
respect to a previous talkspurt not immediately adjacent to the
current talkspurt, and by referring to the description hereinbefore
it is easy to anticipate how to estimate the offset value.
[0073] Again alternatively, in a variation 700 as shown in FIG. 7,
the apparatus for controlling a jitter buffer may further comprise
an offset buffer 702 for storing at least one earlier offset value
of at least one first frame in at least one earlier talkspurt, and
the inter-talkspurt delay jitter estimator 202 may be configured to
adjust the offset value of the delay of the first frame in the
current talkspurt based on an average or weighted average of the at
least one earlier offset value, or adopt the largest offset value
among the present offset value and the at least one earlier offset
value.
[0074] The benefit of using a previous talkspurt not immediately
adjacent to the present talkspurt or more than one previous
talkspurt is to get a more reliable estimate of the offset. The
immediate previous talkspurt might be very short, which makes the
anchor less accurate. Therefore it would make sense to use an
earlier and more reliable talkspurt, or several previous talkspurts
to find the talkspurt with minimum delay (the corresponding offset
would be the maximum).
[0075] However, if we always use the minimum anchor, eventually we
may be able to find out the absolute minimum anchor throughout the
whole conversation session. But this may not be a good solution as
we may likely have a large offset value, and fail to address the
clock drift problem (resetting a new reference anchor frame for
inter-talkspurt delay jitter estimation would implicitly address
the clock drift problem). So we may just use a limited number of
previous talkspurts. The number may be limited with a predefined
threshold, or by a predefined time period. For example, it may be
stipulated that only those talkspurts within last 30 seconds may be
considered.
Estimation of and Adjustment to Long Term Length and Instant Length
of Jitter Buffer
[0076] In the first embodiment discussed hereinbefore, the instant
length of the jitter buffer (JB length) may be adjusted based on
the long term length of the jitter buffer and the offset value as
determined in the first embodiment. Such adjustment may be
implemented in any proper way in so far as both the long term
length and the offset value are considered. For example, as shown
in the second embodiment 800 illustrated in FIG. 8, the JB length
may be firstly determined based on the long term length from a long
term length estimator 802 (e.g. JB length=a*long term length, where
a is a coefficient), then adjusted with the offset value from the
inter-talkspurt delay jitter estimator 202.
[0077] The long term length B.sub.j(n) of the jitter buffer for
frame n in talkspurt j can be estimated by computing a statistical
distribution of history delay jitter values, such as a histogram of
the past jitter values. By making use of the histogram, a delay
jitter threshold may be selected so that the cumulative probability
of the delay jitter values lower than the delay jitter threshold
meets the requirement in practice. For example, if a voice
communication application requires the frame loss rate shall be
lower than 5%, then we can set the delay jitter threshold so that
the cumulative probability of the delay jitter values lower than
the delay jitter threshold is equal or greater than 95%. If the
jitter buffer length is set to be equal to the so-determined
threshold, then the frame loss rate will be equal to or lower than
5% (although in the present application, the long term length is
not necessarily directly adopted as the instant jitter buffer
length). Alternatively, the histogram can be replaced by a
probability mass function (PMF) (see US patent application
published as US20090003369A1 the contents of which are incorporated
by reference in its entirety).
[0078] Note that since the long term length is determined based on
jitter values in the history, the history data will be updated only
when a new frame is received. Therefore, the estimation of the long
term length of the long term length estimator 802 will be triggered
only when a new frame is received. At a particular time, we may or
may not receive a frame to update the long term length. Therefore,
in the symbol of the long term length B.sub.j(n), we ignore the
time index t.
[0079] In a variation 900 of the second embodiment, as shown in
FIG. 9, the long term length B.sub.j(n) may be firstly adjusted by
a long term length adjustor 902 with the offset value d.sub.j(0)
discussed in the first embodiment. The basic idea of the adjustment
is properly decreasing the long term length if the offset value is
greater than 0.
[0080] The decrease of the long term length may be realized in any
way. As an example, the following formula or any variation thereof
may be adopted:
B j ' .function. ( n ) = B j .function. ( n ) + max .function. ( 0
, B j .function. ( n ) - d j .function. ( 0 ) ) 2 ( 1 )
##EQU00001##
where B'.sub.j(n) is an adjusted version of the long term length
B.sub.j(n), and d.sub.j(0) should be not less than 0. The operation
.left brkt-bot..chi..right brkt-bot. has the same meaning as in the
first embodiment, and may be replaced with rounding function or
ceiling function.
[0081] If the offset is greater than the long term length, it means
the jitter has exhausted the long term length, and the packet
should be played out immediately. While in the present application,
according to the formula (1), we still give the packet a smaller
buffer length (B.sub.j(n)/2) to avoid possible packet loss. Here,
the denominator "2" is just an example and it may be any other
value greater than 1.
[0082] When d.sub.j(0) is less than B.sub.j(n), it means the long
term length has been partly "used" by the offset of the first
packet of the new talkspurt, and the adjusted long term length may
be larger than the former case.
[0083] In another variation, the long term length adjustor 902 may
be configured to fade the offset value with time. That is,
d.sub.j(0) calculated in the first embodiment may just serve as an
initial offset used in the formula (1), and with the increasing of
the index n (as well as lapse of time), that is, every time a new
frame is received, the value of d.sub.j(0) may be decreased, thus
the effect of d.sub.j(0) upon the long term length may be faded
with time.
[0084] In yet another variation, it is proposed to discard the
effect of previous talkspurt(s) if the transmission quality for the
previous talkspurt(s) is too bad. That is, if the long term length
at the end of the last talkspurt is greater than a predefined
threshold, for the new talkspurt, the long term length estimator
may reset the long term length to a predefined initial value, and
reset the history delay jitter values. That is, the history jitter
values up to the last talkspurt will be discarded and the history
jitter values will be re-accumulated from the start of the new
talkspurt. In this way, we can reset the long term length (and thus
the length of the jitter buffer) to a normal level, since the last
talkspurt probably is an abnormal one.
[0085] Please note that the solution of the second embodiment may
be based on the first embodiment, but it can also be a solution
independent from the first embodiment. Therefore, as shown in FIG.
8 and FIG. 9, those components shown in dash lines may be
omitted.
Further Adjustment to Initial Length of Jitter Buffer for New
Talkspurt
[0086] In the first and second embodiments discussed hereinbefore,
the jitter buffer length may be adjusted via adjustment to the long
term length of the jitter buffer. A simple way is directly using
the long term length as the instant length of the jitter buffer for
each frame.
[0087] In a third embodiment, a more specific adjustment to the
jitter buffer length, or strictly speaking, the initial length of
the jitter buffer for a new talkspurt, is introduced.
[0088] The third embodiment is also shown in FIG. 8 and FIG. 9,
especially by the dashed line arrow directing from the jitter
buffer 206 to the JB controller 204. Specifically, the jitter
buffer controller 204 may be configured to set the length of the
jitter buffer 206 for the first frame of the current talkspurt
based on the long term length for the first frame and the number of
frames received at the same time with the first frame. For example,
the following formula or any variation thereof may be used:
L.sub.j(t0)=B'.sub.j(n0)+N(t0)+C (2)
where L.sub.j(t0) is the instant length of the jitter buffer at the
time t0 of the first frame n0 in the current talkspurt,
B'.sub.j(n0) is the adjusted long term length for the first frame
n0, N(t0) is the number of frames received at t0, that is, at the
same time as the first frame n0, and C is a predefined constant
integer. Here, the expression "at the same time" shall be construed
as mentioned before. That is, N(t0) actually is the number of
frames received in the same time gap as the first frame n0. Here
N(t0) may or may not take the first frame itself into account, and
the difference so introduced may be compensated by the constant
C.
[0089] In the formula an adjusted version B'.sub.j(n0) of the long
term length is used. That is, the third embodiment is based on the
first and/or second embodiment. In a fourth embodiment independent
from the first and/or second embodiment, the formula (2) may be
rewritten as:
L.sub.j(t0)=B.sub.j(n0)+N(t0)+C (2')
where B.sub.j(n0) is the long term length for the first frame n0,
but without the adjustment with the offset value d.sub.j(0).
[0090] As discussed in the previous embodiments and their
variations, focus is on reset of long term length and/or initial
length of the jitter buffer for the first frame of the current
talkspurt. As the result of the reset (probably as well as the
improvement of the network condition), such a scenario may occur
where the frames of the last talkspurt have not been played out
completely when some frames of the new talkspurt have come. At that
time, if the long term JB length (adjusted or not adjusted) for the
first frame of the current talkspurt is longer than the number of
the not-yet-played out frames of the last talkspurt, since the
first frame of the current talkspurt must wait for a period of time
corresponding to the long term length, during which the frames of
the last talkspurt will have been played out completely. However,
if the number of not-yet-played out frames of the last talkspurt
(that is the jitter buffer length still used by the last talkspurt)
is greater than the long term length for the first frame of the
current talkspurt, it means not all of those frames of the last
talkspurt can be played out if the first frame of the current
talkspurt must be played out in time based on the long term length
for the first frame. Therefore, in a variation of the third/fourth
embodiment, the jitter buffer controller 204 may be configured to,
if the jitter buffer length still used by the last talkspurt is
greater than the long term length and the first frame of the new
talkspurt has arrived, drop some of the frames of the last
talkspurt so that the frames of the last talkspurt occupy only the
long term length of the jitter buffer.
[0091] For example, as shown in FIG. 10, at the time t1, there are
still 5 frames (n-4 to n) of the last talkspurt in the jitter
buffer, that is, the JB length is 5 frames. Then at the next time
t2, 3 frames (1 to 3, including the first frame 1) of a new
talkspurt arrive and the estimated (and possibly adjusted) long
term length of the jitter buffer for the first frame is 2. That
means the long term length (2) for the first frame 1 plus the
number (3) of the arrived frames of the new talkspurt will occupy a
JB length of 5 frames. If without dropping some frames of the last
talkspurt, then the first frame 1 of the current talkspurt must
wait for 4 frames (n-3 to n, and the frame n-4 has been played
out), which deviate from the long term length of 2 frames too much.
Therefore, 2 frames of the last talkspurt must be dropped to let
the first frame of the current talkspurt be played out timely. In
the example as shown in FIG. 10, frames n-1 and n are dropped, that
is, the last frames are dropped. However, other dropping scheme may
be adopted, which will be discussed later.
Intra-Talkspurt Adaptation of Jitter Buffer Length
[0092] According to a fifth embodiment of the present application,
the instant length of the jitter buffer may be adaptively changed
from frame to frame. As a long term goal, the jitter buffer
controller 204 may be further configured to adjust the present
length of the jitter buffer toward the long term length. Such
adjustment may be implemented by calculating a weighted average of
the long term length and the present length. The following formula
or any variation thereof is an example:
L.sub.f,j(n)=L.sub.a,j(n)+.alpha.(B'.sub.j(n)-L.sub.a,j(n)) (3)
where B'.sub.j(n) is the long term length of the jitter buffer for
the current frame n in talkspurt j, which has been adjusted with
the offset value as described in previous embodiments and their
variants, L.sub.a,j(n) is the actual jitter buffer length for the
current frame n, and L.sub.f,j(n) is the final jitter buffer length
for the current frame n, and .alpha. is the weight, which may be
regarded as a time constant controlling how much smoothing is
applied.
[0093] In the above embodiment, the index n is used because the
operation is executed from frame to frame. In a variation of the
embodiment, the operation may also be executed at regular time
interval, usually the playout tick, rather than on a
per-received-frame basis. Therefore, the formula (3) may be
rewritten as:
L.sub.f,j(t)=L.sub.a,j(t)+.alpha.(B'.sub.j(n)-L.sub.a,j(t))
(3')
where B'.sub.j(n) is the long term length of the jitter buffer for
the current frame n in talkspurt j, which has been adjusted with
the offset value as described in previous embodiments and their
variants, L.sub.a,j(t) is the actual jitter buffer length for the
current time t, and L.sub.f,j(t) is the final jitter buffer length
for the current time t, and .alpha. is the weight, which may be
regarded as a time constant controlling how much smoothing is
applied. Note that the index n is still used in the term
B'.sub.j(n) because B'.sub.j(n) is updated on a frame-basis as
discussed hereinbefore. Also please note that in formula (3') the
current time t does not necessarily correspond to the current frame
n in a one-to-one manner, because the time t is always ticking
forward while the index n will not necessarily increase so smoothly
due to the delay jitter of frames.
[0094] In another variation, we can adopt different adaptation rate
depending on whether the instant buffer length is greater or
smaller than the long term estimate. And the jitter buffer
controller may be configured so that the weight of the long term
length is greater when the long term length is greater than the
present length, and is smaller when the long term length is smaller
than the present length.
[0095] In the fifth embodiment and its variations discussed above,
B'.sub.j(n) is used. However, B.sub.j(n) may be used instead.
Furthermore, the initial value of L.sub.a,j(t) in the current
talkspurt, that is L.sub.a,j(t0), may or may not adopt the value
L.sub.j(t0) discussed in the third/fourth embodiment and their
variants. That is to say, the fifth embodiment and its variations
may be combined together with the other embodiments and their
variations discussed in this application, and may also be
independent therefrom.
Pre-Emptive Expansion of Jitter Buffer
[0096] Jitter is the delay fluctuation of the arrived frames. The
embodiments and variations discussed hereinbefore predict the
jitter buffer length based on jitters of past frames. When there is
a large jitter, the buffer algorithm cannot be updated
appropriately since the instantaneous jitter is not yet seen by the
buffer algorithm. Without adjusting the JB length, a large jitter
will cause a large amount of frame losses.
[0097] So, in a sixth embodiment shown in FIG. 11, it is proposed
to perform pre-emptive buffer expansion based on monitoring the
"future" of the jitter buffer. Specifically, the apparatus 1100 for
controlling a jitter buffer 206 may comprise a jitter buffer
monitor 1102 for monitoring occupancy of the jitter buffer 206, and
wherein the jitter buffer controller 204 is further configured to
increase the length of the jitter buffer 206 in response to the
occupancy meeting a predefined condition.
[0098] The idea is before the buffer getting completely dry/empty,
we pre-emptively increase jitter buffer length by inserting
frames.
[0099] One embodiment is to check the occurrence number of
consecutive single frame buffer occupancy. For example when there
are two consecutive single-frame buffer occupancies, it may
indicate that a rise of jitter may have occurred and the current
buffer length is not long enough to meet the requirement. Therefore
an extra one frame delay may be added to the current buffer
length.
[0100] To be generalized, the jitter buffer controller 204 may be
configured to increase the length of the jitter buffer 206 where an
occupancy less than a predetermined threshold has consecutively
occurred for a predetermined times.
[0101] Since pre-emptive expansion occurs when there are still
frames available in the buffer, it provides the benefit of using
frame interpolation instead of single-side frame extrapolation as
in conventional approaches. It is known to those skilled in the art
that frame interpolation can often produce better result than frame
extrapolation. Pre-emptive expansion also helps spread the signal
process which often causes distortion in the output.
[0102] The jitter buffer length cannot be increased too fast and/or
unlimitedly. In a variation, the jitter buffer controller 204 may
be configured to increase the length of the jitter buffer by one
frame each time the condition is met. In another variation, the
apparatus for controlling a jitter buffer may further comprise a
signal content monitor 1104 for monitoring the content of the audio
signal, and wherein the jitter buffer controller 204 is configured
not to increase the length of the jitter buffer where the content
of the audio signal is not appropriate for frame insertion.
[0103] A specific example is transient signal area. Frame insertion
at the transient signal area often generates artifacts. Therefore,
the signal content monitor 1104 may be configured to monitor the
transient state of the audio signal, and the jitter buffer
controller 204 may be configured not to increase the length of the
jitter buffer in transient signal area.
[0104] Furthermore, the current jitter buffer length cannot be too
long. In a variation, the jitter buffer controller may be
configured to not increase the length of the jitter buffer where
the present jitter buffer length exceeds a predetermined
threshold.
[0105] Past signal processing frequency may also be referred to,
i.e. we may stop frame insertion if this has been done a few times
already. Therefore, in a variation 1200 as shown in FIG. 12, the
apparatus for controlling a jitter buffer may further comprise a
counter 1202 for counting frame insertions within the present
talkspurt, wherein the jitter buffer controller 204 may be
configured to not increase the length of the jitter buffer where
the number of frame insertions within the present talkspurt exceeds
a predetermined threshold.
[0106] Note that similar to other embodiments, the sixth embodiment
and its variations may be implemented on the basis of any one or
more of the other embodiments discussed hereinbefore, or may be
implemented independent from the other embodiments. Just as shown
in FIG. 11 and FIG. 12, all the components shown in dashed lines
may be omitted or may be combined with the present embodiment in
any combination.
Buffer Operation
[0107] In the previous embodiment and their variations, the instant
jitter buffer length may be increased or shrunk in various
situations. When the target jitter buffer length (the calculated
jitter buffer length) equals to the present actual buffer length,
the head frame of the jitter buffer is pushed out and played. In
some other situations, for example where a pre-emptive expansion
command has been issued, or if the target jitter buffer length is
greater than the actual buffer length, then an expansion is
performed. On the other hand, if the target jitter buffer length is
smaller than the actual buffer length, a buffer shrink is
performed.
[0108] Increase of the jitter buffer length may be realized through
frame interpolation when a frame is available in the jitter buffer,
or frame extrapolation when no frames are available in the jitter
buffer. Frame interpolation means adding a frame before an
available frame in the jitter buffer. The added frame may be an
empty frame, a synthesized frame or a duplicate of the present
playout frame or a future (not yet played out) frame in the jitter
buffer. Frame extrapolation means adding a frame at the end of the
jitter buffer. When there is no available frame in the jitter
buffer, it means the added frame is after the latest playout frame.
Similar to interpolation, an extrapolated frame may also be an
empty frame, a synthesized frame or a duplicate of the latest
playout frame.
[0109] Decrease of the jitter buffer length may be realized through
dropping frames in the jitter buffer. The dropped frame may be at
the head or end of the jitter buffer, or in the middle of the
jitter buffer. When considering the position of frames in a
talkspurt, generally we would drop ending frames in the jitter
buffer when the talkspurt is coming to an end. When it is in the
middle of a talkspurt, then we may drop earlier frames in the
jitter buffer.
[0110] For avoiding possible artifacts originating from consecutive
frames dropping, the jitter buffer controller may be configured to
decrease the jitter buffer length through interleaved removal of
frames in the jitter buffer. The removed frames may be selected
randomly or according to predetermined pattern, such as removing
one frame from every two or three frames.
[0111] Or, the apparatus for controlling the jitter buffer may
further comprise a frame evaluator for evaluating the importance of
frames in the jitter buffer, wherein the jitter buffer controller
is configured to decrease the jitter buffer length by dropping
those least important frames. The importance of the frame may be
evaluated based on signal energy or spectral flux of frames, those
frames with relatively low signal energy or low spectral flux being
determined as relatively less important frames.
[0112] Another factor that may be used as the standard for dropping
frames is periodicity. If the signal is highly periodic, it will
make time scaling (expansion or compression) less prone to
artifacts. Therefore, those segments of an audio signal that are
highly periodic may be subject to interleaved removal; while the
other segments may be subject to other form of frame dropping if
necessary, such as dropping frames of lower energy.
Combination of Embodiments and Application Scenarios
[0113] All the embodiments and variants there of discussed above
may be implemented in any combination thereof, and any components
mentioned in different parts/embodiments but having the same or
similar functions may be implemented as the same or separate
components.
[0114] Specifically, when describing the embodiments and their
variations hereinbefore, those components having reference signs
similar to those already described in previous embodiments or
variants are omitted, and just different components are described.
In fact, these different components can either be combined with the
components of other embodiments or variants, or constitute separate
solutions alone. For example, there may be the following separate
solutions: (1) using inter-talkspurt delay jitter (offset value) to
adjust jitter buffer length (FIG. 2); (2) setting anchor frame
(FIG. 4); (3) estimating intra-talkspurt delay jitter and updating
anchor frame (FIG. 3, FIG. 5); (4) estimating inter-talkspurt delay
jitter (offset value) (FIG. 6); (5) using multiple offset values to
adjust the jitter buffer length (FIG. 7); (6) estimating long term
jitter buffer length (FIG. 8); (7) determining the jitter buffer
length based on long term jitter buffer length and the offset value
(FIG. 8); (8) setting and adapting instant jitter buffer length
based on the long term jitter buffer length and first received
frames (FIG. 10); (9) adjusting the long term jitter buffer length
with the offset value (FIG. 9); (10) intra-talkspurt adaption of
the jitter buffer length toward the long term jitter buffer length;
(11) pre-emptive expansion of the jitter buffer; and (12)
expansion/increase and shrinking/decrease of the jitter buffer.
And, any two or more of the above separate solutions may be
combined together to form new solutions.
[0115] As discussed at the beginning of the Detailed Description of
the present application, the embodiment of the application may be
embodied either in hardware or in software, or in both. FIG. 13 is
a block diagram illustrating an exemplary system for implementing
the aspects of the present application.
[0116] In FIG. 13, a central processing unit (CPU) 1301 performs
various processes in accordance with a program stored in a read
only memory (ROM) 1302 or a program loaded from a storage section
1308 to a random access memory (RAM) 1303. In the RAM 1303, data
required when the CPU 1301 performs the various processes or the
like are also stored as required.
[0117] The CPU 1301, the ROM 1302 and the RAM 1303 are connected to
one another via a bus 1304. An input/output interface 1305 is also
connected to the bus 1304.
[0118] The following components are connected to the input/output
interface 1305: an input section 1306 including a keyboard, a
mouse, or the like; an output section 1307 including a display such
as a cathode ray tube (CRT), a liquid crystal display (LCD), or the
like, and a loudspeaker or the like; the storage section 1308
including a hard disk or the like; and a communication section 1309
including a network interface card such as a LAN card, a modem, or
the like. The communication section 1309 performs a communication
process via the network such as the internet.
[0119] A drive 1310 is also connected to the input/output interface
1305 as required. A removable medium 1311, such as a magnetic disk,
an optical disk, a magneto-optical disk, a semiconductor memory, or
the like, is mounted on the drive 1310 as required, so that a
computer program read there from is installed into the storage
section 1308 as required.
[0120] In the case where the above-described components are
implemented by the software, the program that constitutes the
software is installed from the network such as the internet or the
storage medium such as the removable medium 1311.
Method for Controlling a Jitter Buffer
[0121] In the process of describing the apparatus for controlling a
jitter buffer in the embodiments hereinbefore, apparently disclosed
are also some processes or methods. Hereinafter a summary of these
methods is given without repeating some of the details already
discussed hereinbefore, but it shall be noted that although the
methods are disclosed in the process of describing the apparatus
for controlling a jitter buffer, the methods do not necessarily
adopt those components as described or are not necessarily executed
by those components. For example, the embodiments of the apparatus
for controlling a jitter buffer may be realized partially or
completely with hardware and/or firmware, while it is possible that
the method for controlling a jitter buffers discussed below may be
realized totally by a computer-executable program, although the
methods may also adopt the hardware and/or firmware of the
apparatus for controlling a jitter buffer.
[0122] The methods will be described below with reference to FIGS.
14-18. Please note that in correspondence to the streaming property
of the audio signal, the control of jitter buffer is a
time-dependent process. For different aspects of the control, there
are different threads. For example but not intended to be
limitative, FIGS. 14-18 show several threads among others. FIG. 14
is a thread 1400 for setting and updating of anchor frames. FIG. 15
is a thread 1500 for estimating the long term length of the jitter
buffer. FIG. 16 is a thread 1600 for estimating the cross-talkspurt
offset and adjusting the long term length of the jitter buffer.
FIG. 17 is a thread 1700 for updating the jitter buffer length. And
FIG. 18 is a thread 1800 for pre-emptively expanding the jitter
buffer. In each thread, some operations, such as the operation of
estimating, may comprise more operations which however are not
necessarily shown in FIG. 14-18. In addition, the results of some
of the threads will be used by some other threads. Therefore, the
embodiments of the methods for controlling a jitter buffer
according to the present application may be only a part of one of
the threads shown in FIGS. 14-18, or may be a solution bridging
different parts of different threads. For those details not
illustrated in FIGS. 14-18 and/or not described below, reference
may be made to the description of the apparatus for controlling a
jitter buffer.
[0123] According to a seventh embodiment of the present
application, a method for controlling a jitter buffer is provided.
According to the method, an offset value of the delay of a first
frame in the current talkspurt is estimated with respect to the
delay of a latest anchor frame in a previous talkspurt, and is used
to, together with a long term length of the jitter buffer for each
frame, to adjust a length of the jitter buffer. A positive offset
value means the first frame in a new talkspurt probably is delayed
too much, that is, has a positive jitter. In such a situation,
depending on requirements in practice, the long term length of the
jitter buffer may be increased to accommodate possibly greater
jitter, or may be decreased to avoid the delay introduced by the
jitter buffer is too long.
[0124] FIG. 14 shows the setting of the anchor frame. Initially,
that is when receiving a new frame (operation 1402) and the new
frame is a first frame in a new talkspurt ("Yes" in operation
1404), then the frame is set as anchor frame (operation 1406).
Then, newly received frame after the first frame ("No" in operation
1404) may be set as the anchor frame if it meets a predefined
condition ("Yes" in operation 1408). That is, the anchor frame may
be updated to be the newly received frame.
[0125] The predefined condition may be about the delay jitter of
the newly received frame. If the delay jitter is less than 0, or
equal to 0, then the newly received frame may be set as new anchor
frame. In this way, it is ensured that finally a local minimum
delay may be located within each talkspurt, thus avoiding the
jitter buffer length becomes too long. The delay jitter may be
calculated as a difference between an actual reception time of the
newly received frame and an expected reception time of the same
newly received frame. The expected reception time is calculated
with the present anchor frame as a reference.
[0126] The offset value mentioned above may be estimated with
respect to the latest anchor frame in a previous talkspurt
immediately previous to the current talkspurt, or a previous
talkspurt not immediately adjacent to the current talkspurt. In
this way, some previous talkspurts which are too short or too bad
in quality may be neglected, and, similar to the estimation of the
intra-spurt delay jitter, it can be ensured that the reference
(anchor frame in a previous talkspurt) itself can have relatively
small jitter.
[0127] Alternatively, when estimating the offset value, the offset
value of the delay of the first frame in the current talkspurt may
be adjusted based on an average or weighted average of at least one
earlier offset value, or adopt the largest offset value among the
present offset value and the at least one earlier offset value.
[0128] The offset value may be calculated as a time difference
between the following two: a real time difference between the
reception time of the latest anchor frame and the reception time of
the first frame, and an corresponding expected time difference. The
expected time difference may be obtained based on an expected
number of frames between the latest anchor frame and the first
frame.
[0129] For getting the expected number of frames, the expected
number of frames may be obtained firstly based on sequence numbers
of the latest anchor frame and the first frame. Also, information
regarding silence frames between the previous talkspurt and the
current talkspurt may be retrieved from at least one frame in at
least one talkspurt in the previous talkspurt and the current
talkspurt.
[0130] For example, the information regarding silence frames may
include the number of the silence frames, the number being embedded
in the first frame of the current talkspurt. Or, the information
regarding silence frames may comprise timestamps in the last frame
of the previous talkspurt and the first frame in the current
talkspurt.
[0131] According to an eighth embodiment of the application, the
method may further comprise estimating the long term length for
each frame by calculating a statistic distribution of history delay
jitter values (operation 1512 or operation 1514 in FIG. 15), such
as a histogram or a probability mass function (PMF) of the past
jitter values. Since the long term length is calculated based on
delay jitter values of past frames, the calculation or updating
thereof may be triggered by reception of a new frame (operation
1502 in FIG. 15).
[0132] As a variation, for avoiding bad quality of a previous
talkspurt increases the long term length (and thus the instant
jitter buffer length) too much, an opportunity of cutting down too
long a jitter buffer is provided upon the start of a new talkspurt
("Yes" in operation 1504). That is, when the long term length at
the end of the last talkspurt is greater than a predefined
threshold Th1 ("Yes" in operation 1508), the long term length may
be reset to an initial value and the history delay jitter values
may be emptied (operation 1510). The history data will be
re-cumulated from the first frame, and from the first frame on, the
long term length will be estimated based on history jitter values
since the first frame (operation 1512). If the long term length at
the end of the last talkspurt is not greater than a predefined
threshold Th1 ("No" in operation 1508), the long term length will
not be reset and will be continued to be estimated based on history
jitter values since beginning or the last resetting.
[0133] In a variation 1600 as shown in FIG. 16, the long term
length may be subject to an adjustment based on the offset value,
and all the operations relating to the long term length discussed
above are based on the adjusted value of the long term length. For
example, when the offset value is greater than zero ("Yes" in
operation 1608), the long term length may be decreased (operation
1610). In FIG. 16 it is shown that no operations will be performed
when the offset is not greater than 0, but the present application
is not limited thereto. For example, depending on requirements in
practice, sometimes it may be acceptable to increase the long term
length when the offset is not greater than 0.
[0134] It can be recalled that the offset value is estimated
(operation 1606) for the first frame in a new talkspurt ("Yes" in
operation 1604). When receiving a new frame (operation 1602)
subsequent to the first frame ("No" in operation 1604), the
corresponding long term length for each frame may also be adjusted
(operation 1610). As a further variation, for each subsequent
adjustment to the long term length, the offset value may fade
gradually (operation 1612), since the farther away is the
subsequent frame from the first frame, the smaller is the effect of
the offset of the first frame.
[0135] In a ninth embodiment of the application, the operation of
adjusting the length of the jitter buffer may comprise, for the
first frame of the current talkspurt (operation 1702 in FIG. 17)
setting the length of the jitter buffer based on the long term
length for the first frame and the number of frames received at the
same time with the first frame (operation 1708), so as to ensuring
the delay time of the first frame and the buffering space for the
other frames already received.
[0136] In such a situation, where the jitter buffer length still
used by the last talkspurt (that is the present actual jitter
buffer length) is greater than the long term length ("Yes" in
operation 1704) and the first frame of the new talkspurt has
arrived (operation 1802), some of the frames of the last talkspurt
may be dropped (operation 1706) so that the frames of the last
talkspurt occupy only the long term length of the jitter buffer. If
the present jitter buffer length is not greater than the long term
length ("No" in operation 1704), then the frames (if any) of the
last talkspurt have enough time to be played out before the first
frame of the new talkspurt is played out. The excessive time will
be occupied by empty frames in the jitter buffer before the first
frame.
[0137] With the lapse of time ("Yes" in operation 1712) or upon
receiving a new frame ("Yes" in operation 1710), the jitter buffer
length may be further adjusted toward the long term length
(operation 1714). Here, the "time interval" may be the "processing
time granularity" as mentioned before, and the word "toward" means
the jitter buffer length is adjusted to be closer to the long term
length, but not necessarily become equal to the long term length at
once.
[0138] To adjust the present length of the jitter buffer toward the
long term length, we can calculate a weighted average of the long
term length and the present length. When the long term length is
greater than the present length, the weight of the long term length
may be greater so that the jitter buffer length may be increased
quickly to the long term length because the situation where the
long term length is great generally means the delay is severe and
longer jitter buffer is needed to prevent frame loss. On the
contrary, when the long term length is smaller than the present
length, the weight may be smaller, because in such a situation, the
purpose of the adjustment is to release unnecessary length of the
jitter buffer so as to shorten the delay time, which is not so
urgent as preventing frame loss. Nevertheless, sometimes we may
directly set the long term length as the length of the jitter
buffer for each frame.
[0139] According to a tenth embodiment of the application, the
method may further comprise pre-emptively expanding the jitter
buffer (operation 1718 in FIG. 17) where necessary ("Yes" in
operation 1716), so as to make ready for possible rise of delay
jitter.
[0140] Such possible rise of delay jitter may be reflected by
continuous low occupancy of the jitter buffer, meaning that the
receiving of future frames are becoming slower. Therefore, we may
monitor the occupancy of the jitter buffer (operation 1802 in FIG.
18), and increase the length of the jitter buffer (operation 1814)
in response to the occupancy meeting a predefined condition, such
as continuous low occupancy ("Yes" in operation 1804). In a
variation, the predefined condition is the occupancy is less than a
predetermined threshold consecutively for a predetermined times,
such as a single-frame occupancy has consecutively occurred
twice.
[0141] Since the pre-emptive expansion is based on a prediction, it
is better to be cautious so that the jitter buffer length will not
be increased too much and too quickly. Therefore we may pose some
limitations on the pre-emptive expansion of the jitter buffer.
[0142] In one variation, we may monitor the content of the audio
signal (operation 1806), and do not increase the length of the
jitter buffer if the content of the audio signal is not appropriate
for frame insertion ("No" in operation 1808). For example, we may
monitor the transient state of the audio signal, and do not
increase the length of the jitter buffer in transient signal area
because in such area insertion of frame(s) will cause remarkable
artifacts.
[0143] Further, if the present jitter buffer length is already
long, then it is not proper to further increase the jitter buffer
length. So we can set a predetermined threshold Th2, beyond which
the operation of increasing the length of the jitter buffer will
not be performed ("Yes" in operation 1810).
[0144] Another criterion is times of frame insertions. We can count
frame insertions within the present talkspurt (operation 1816).
When the number of frame insertions within the present talkspurt
exceeds a predetermined threshold Th3 ("Yes" in operation 1812),
the operation of increasing the length of the jitter buffer will
not be performed.
[0145] Similar to the embodiments of the apparatus for controlling
a jitter buffer, any combination of the seventh to tenth embodiment
and their variations are practical on one hand; and on the other
hand, every aspect of the embodiments and their variations may be
separate solutions.
[0146] Please note the terminology used herein is for the purpose
of describing particular embodiments only and is not intended to be
limiting of the application. As used herein, the singular forms
"a", "an" and "the" are intended to include the plural forms as
well, unless the context clearly indicates otherwise. It will be
further understood that the terms "comprises" and/or "comprising,"
when used in this specification, specify the presence of stated
features, integers, operations, steps, elements, and/or components,
but do not preclude the presence or addition of one or more other
features, integers, operations, steps, elements, components, and/or
groups thereof.
[0147] The corresponding structures, materials, acts, and
equivalents of all means or operation plus function elements in the
claims below are intended to include any structure, material, or
act for performing the function in combination with other claimed
elements as specifically claimed. The description of the present
application has been presented for purposes of illustration and
description, but is not intended to be exhaustive or limited to the
application in the form disclosed. Many modifications and
variations will be apparent to those of ordinary skill in the art
without departing from the scope and spirit of the application. The
embodiment was chosen and described in order to best explain the
principles of the application and the practical application, and to
enable others of ordinary skill in the art to understand the
application for various embodiments with various modifications as
are suited to the particular use contemplated.
[0148] As will be apparent to those skilled in the art in the light
of the foregoing disclosure, many alterations and modifications are
possible in the practice of this invention without departing from
the spirit or scope thereof. Accordingly, the scope of the
invention is to be construed in accordance with the substance
defined by the following claims.
[0149] Accordingly the invention may suitably comprise, consist of,
or consist essentially of, any of element (the various parts or
features of the invention and their equivalents as described
herein, currently existing, and/or as subsequently developed.
Further, the present invention illustratively disclosed herein may
be practiced in the absence of any element, whether or not
specifically disclosed herein. Obviously, numerous modifications
and variations of the present invention are possible in light of
the above teachings. It is therefore to be understood that within
the scope of the appended claims, the invention may be practiced
otherwise than as specifically described herein.
[0150] Accordingly, the invention may be embodied in any of the
forms described herein, including, but not limited to the following
Enumerated Example Embodiments (EEEs) which described structure,
features, and functionality of some portions of the present
invention.
[0151] EEE1. An apparatus for controlling a jitter buffer
comprising: [0152] an inter-talkspurt delay jitter estimator for
estimating an offset value of the delay of a first frame in the
current talkspurt with respect to the delay of a latest anchor
frame in a previous talkspurt; and [0153] a jitter buffer
controller for adjusting a length of the jitter buffer based on a
long term length of the jitter buffer for each frame and the offset
value.
[0154] EEE2. The apparatus for controlling a jitter buffer
according to EEE 1, further comprising an anchor frame setting unit
for initially setting the first frame in the current talkspurt as
the anchor frame, and setting a newly received frame as the anchor
frame when a delay jitter of the newly received frame meets a
predefined condition.
[0155] EEE3. The apparatus for controlling a jitter buffer
according to EEE 2, wherein the predefined condition is the delay
jitter is less than 0.
[0156] EEE4. The apparatus for controlling a jitter buffer
according to EEE 2, wherein the predefined condition is the delay
jitter is equal to or less than 0.
[0157] EEE5. The apparatus for controlling a jitter buffer
according any one of EEE's 1-4, further comprising an
intra-talkspurt delay jitter estimator for calculating, as the
delay jitter, a difference between an actual reception time of the
newly received frame and an expected reception time of the same
newly received frame, the expected reception time being calculated
with the present anchor frame as a reference.
[0158] EEE6. The apparatus for controlling a jitter buffer
according to any one of EEE's 1-5, wherein the inter-talkspurt
delay jitter estimator is configured to estimate the offset value
of the delay of the first frame in the current talkspurt with
respect to the delay of the latest anchor frame in a previous
talkspurt immediately previous to the current talkspurt.
[0159] EEE7. The apparatus for controlling a jitter buffer
according to EEE 6, wherein the inter-talkspurt delay jitter
estimator comprises: [0160] a time difference estimator for
calculating a time difference between the reception time of the
latest anchor frame and the reception time of the first frame, and
[0161] a frame number estimator for estimating an expected number
of frames between the latest anchor frame and the first frame; and
wherein [0162] the inter-talkspurt delay jitter estimator is
configured to estimate the offset value based on the time
difference and an expected time difference obtained based on the
expected number of frames.
[0163] EEE8. The apparatus for controlling a jitter buffer
according to EEE 7, wherein the frame number estimator is
configured to obtain the expected number of frames based on
sequence numbers of the latest anchor frame and the first frame and
information regarding silence frames between the previous talkspurt
and the current talkspurt, the information regarding silence frames
being carried in at least one frame in at least one talkspurt in
the previous talkspurt and the current talkspurt.
[0164] EEE9. The apparatus for controlling a jitter buffer
according to EEE 8, wherein the information regarding silence
frames comprises the number of the silence frames, the number being
embedded in the first frame of the current talkspurt.
[0165] EEE10. The apparatus for controlling a jitter buffer
according to EEE 8, wherein the information regarding silence
frames comprises timestamps in the last frame of the previous
talkspurt and the first frame in the current talkspurt.
[0166] EEE11. The apparatus for controlling a jitter buffer
according to any one of EEE;s 1-10, further comprising an offset
buffer for storing at least one earlier offset value of at least
one first frame in at least one earlier talkspurt, wherein the
inter-talkspurt delay jitter estimator is configured to adjust the
offset value of the delay of the first frame in the current
talkspurt based on an average or weighted average of the at least
one earlier offset value, or adopt the largest offset value among
the present offset value and the at least one earlier offset
value.
[0167] EEE12. The apparatus for controlling a jitter buffer
according to any one of EEE's 1-11, further comprising a long term
length estimator for estimating the long term length for each frame
by calculating a statistical distribution of history delay jitter
values.
[0168] EEE13. The apparatus for controlling a jitter buffer
according to any one of EEE's 1-12, further comprising a long term
length adjustor for adjusting the long term length for each frame
to a smaller value when the offset value is greater than zero.
[0169] EEE14. The apparatus for controlling a jitter buffer
according to EEE 13, wherein the long term length adjustor is
configured to fade the offset value with time.
[0170] EEE15. The apparatus for controlling a jitter buffer
according to any one of EEE's 12-14, wherein the long term length
estimator is configured to reset the long term length to an initial
value and reset the history delay jitter values when the long term
length at the end of the last talkspurt is greater than a
predefined threshold.
[0171] EEE16. The apparatus for controlling a jitter buffer
according to any one of EEE's 1-15, wherein the jitter buffer
controller is configured to set the length of the jitter buffer for
the first frame of the current talkspurt based on the long term
length for the first frame and the number of frames received at the
same time with the first frame.
[0172] EEE17. The apparatus for controlling a jitter buffer
according to any one of EEE's 1-16, wherein, the jitter buffer
controller is configured to, if the jitter buffer length still used
by the last talkspurt is greater than the long term length and the
first frame of the new talkspurt has arrived, drop some of the
frames of the last talkspurt so that the frames of the last
talkspurt occupy only the long term length of the jitter
buffer.
[0173] EEE18. The apparatus for controlling a jitter buffer
according to any one of EEE's 1-17, wherein the jitter buffer
controller is further configured to adjust the present length of
the jitter buffer toward the long term length.
[0174] EEE19. The apparatus for controlling a jitter buffer
according to EEE 18, wherein the jitter buffer controller is
configured to adjust the present length of the jitter buffer by
calculating a weighted average of the long term length and the
present length.
[0175] EEE20. The apparatus for controlling a jitter buffer
according to EEE 19, wherein the weight of the long term length is
greater when the long term length is greater than the present
length, and is smaller when the long term length is smaller than
the present length.
[0176] EEE21. The apparatus for controlling a jitter buffer
according to any one of EEE's 1-20, wherein the jitter buffer
controller is configured to determine the length of the jitter
buffer for each frame as the long term length.
[0177] EEE22. An apparatus for controlling a jitter buffer
comprising: [0178] a long term length estimator for estimating a
long term length of the jitter buffer for each frame by calculating
a statistic distribution of history delay jitter values; and [0179]
a jitter buffer controller for setting the length of the jitter
buffer; and wherein [0180] the jitter buffer controller is
configured to set the length of the jitter buffer for the first
frame of the current talkspurt based on the long term length for
the first frame and the number of frames received at the same time
with the first frame.
[0181] EEE23. The apparatus for controlling a jitter buffer
according to EEE 22, wherein, the jitter buffer controller is
configured to, if the jitter buffer length still used by the last
talkspurt is greater than the long term length and the first frame
of the new talkspurt has arrived, drop some of the frames of the
last talkspurt so that the frames of the last talkspurt occupy only
the long term length of the jitter buffer.
[0182] EEE24. The apparatus for controlling a jitter buffer
according to EEE22 or EEE23, wherein the jitter buffer controller
is further configured to adjust the present length of the jitter
buffer toward the long term length.
[0183] EEE25. The apparatus for controlling a jitter buffer
according to EEE 24, wherein the jitter buffer controller is
configured to adjust the present length of the jitter buffer by
calculating a weighted average of the long term length and the
present length.
[0184] EEE26. The apparatus for controlling a jitter buffer
according to EEE 25, wherein the weight of the long term length is
greater when the long term length is greater than the present
length, and is smaller when the long term length is smaller than
the present length.
[0185] EEE27. An apparatus for controlling a jitter buffer
comprising: [0186] a long term length estimator for estimating a
long term length of the jitter buffer for each frame by calculating
a statistic distribution of history delay jitter values; and [0187]
a jitter buffer controller for adjusting the present length of the
jitter buffer toward the long term length.
[0188] EEE28. The apparatus for controlling a jitter buffer
according to Claim 27, wherein the jitter buffer controller is
configured to adjust the present length of the jitter buffer by
calculating a weighted average of the long term length and the
present length.
[0189] EEE29. The apparatus for controlling a jitter buffer
according to EEE 28, wherein the weight of the long term length is
greater when the long term length is greater than the present
length, and is smaller when the long term length is smaller than
the present length.
[0190] EEE30. The apparatus for controlling a jitter buffer
according to any one of EEE's 1-29, further comprising a jitter
buffer monitor for monitoring occupancy of the jitter buffer, and
wherein the jitter buffer controller is further configured to
increase the length of the jitter buffer in response to the
occupancy meeting a predefined condition.
[0191] EEE31. The apparatus for controlling a jitter buffer
according to EEE 30, wherein the jitter buffer controller is
configured to increase the length of the jitter buffer where an
occupancy less than a predetermined threshold has consecutively
occurred for a predetermined times.
[0192] EEE32. The apparatus for controlling a jitter buffer
according to EEE 31, wherein the jitter buffer controller is
configured to increase the length of the jitter buffer where a
single-frame occupancy has consecutively occurred twice.
[0193] EEE33. The apparatus for controlling a jitter buffer
according to any one of EEE's 30-32, wherein the jitter buffer
controller is configured to increase the length of the jitter
buffer by one frame each time.
[0194] EEE34. The apparatus for controlling a jitter buffer
according to any one of EEE's 30-32, further comprising a signal
content monitor for monitoring the content of the audio signal, and
wherein the jitter buffer controller is configured not to increase
the length of the jitter buffer where the content of the audio
signal is not appropriate for frame insertion.
[0195] EEE35. The apparatus for controlling a jitter buffer
according to EEE 34, wherein the signal content monitor is
configured to monitor the transient state of the audio signal, and
the jitter buffer controller is configured not to increase the
length of the jitter buffer in transient signal area.
[0196] EEE36. The apparatus for controlling a jitter buffer
according to any one of EEE's 30-32, wherein the jitter buffer
controller is configured to not increase the length of the jitter
buffer where the present jitter buffer length exceeds a
predetermined threshold.
[0197] EEE37. The apparatus for controlling a jitter buffer
according to any one of EEE's 30-32, further comprising a counter
for counting frame insertions within the present talkspurt, wherein
the jitter buffer controller is configured to not increase the
length of the jitter buffer where the number of frame insertions
within the present talkspurt exceeds a predetermined threshold.
[0198] EEE38. An apparatus for controlling a jitter buffer,
comprising: [0199] a jitter buffer monitor for monitoring occupancy
of the jitter buffer, and [0200] a jitter buffer controller for
increasing the length of the jitter buffer in response to the
occupancy meeting a predefined condition.
[0201] EEE39. The apparatus for controlling a jitter buffer
according to EEE 38, wherein the jitter buffer controller is
configured to increase the length of the jitter buffer where an
occupancy less than a predetermined threshold has consecutively
occurred for a predetermined times.
[0202] EEE40. The apparatus for controlling a jitter buffer
according to EEE 39, wherein the jitter buffer controller is
configured to increase the length of the jitter buffer where a
single-frame occupancy has consecutively occurred twice.
[0203] EEE41. The apparatus for controlling a jitter buffer
according to any one of EEE's 38-40, wherein the jitter buffer
controller is configured to increase the length of the jitter
buffer by one frame each time.
[0204] EEE42. The apparatus for controlling a jitter buffer
according to any one of EEE's 38-40, further comprising a signal
content monitor for monitoring the content of the audio signal, and
wherein the jitter buffer controller is configured not to increase
the length of the jitter buffer where the content of the audio
signal is not appropriate for frame insertion.
[0205] EEE43. The apparatus for controlling a jitter buffer
according to EEE 42, wherein the signal content monitor is
configured to monitor the transient state of the audio signal, and
the jitter buffer controller is configured not to increase the
length of the jitter buffer in transient signal area.
[0206] EEE44. The apparatus for controlling a jitter buffer
according to any one of EEE's 38-40, wherein the jitter buffer
controller is configured to not increase the length of the jitter
buffer where the present jitter buffer length exceeds a
predetermined threshold.
[0207] EEE45. The apparatus for controlling a jitter buffer
according to any one of EEE's 38-40, further comprising a counter
for counting frame insertions within the present talkspurt, wherein
the jitter buffer controller is configured to not increase the
length of the jitter buffer where the number of frame insertions
within the present talkspurt exceeds a predetermined threshold.
[0208] EEE46. A method for controlling a jitter buffer comprising:
[0209] estimating an offset value of the delay of a first frame in
the current talkspurt with respect to the delay of a latest anchor
frame in a previous talkspurt; and [0210] adjusting a length of the
jitter buffer based on a long term length of the jitter buffer for
each frame and the offset value.
[0211] EEE47. The method for controlling a jitter buffer according
to Claim 46, further comprising initially setting the first frame
in the current talkspurt as the anchor frame, and setting a newly
received frame as the anchor frame when a delay jitter of the newly
received frame meets a predefined condition.
[0212] EEE48. The method for controlling a jitter buffer according
to EEE 47, wherein the predefined condition is the delay jitter is
less than 0.
[0213] EEE49. The method for controlling a jitter buffer according
to EEE 47, wherein the predefined condition is the delay jitter is
equal to or less than 0.
[0214] EEE50. The method for controlling a jitter buffer according
any one of EEE's 46-49, further comprising calculating, as the
delay jitter, a difference between an actual reception time of the
newly received frame and an expected reception time of the same
newly received frame, the expected reception time being calculated
with the present anchor frame as a reference.
[0215] EEE51. The method for controlling a jitter buffer according
to any one of EEE's 46-50, wherein the operation of estimating the
offset value comprises estimating the offset value of the delay of
the first frame in the current talkspurt with respect to the delay
of the latest anchor frame in a previous talkspurt immediately
previous to the current talkspurt.
[0216] EEE52. The method for controlling a jitter buffer according
to EEE 51, wherein the operation of estimating the offset value
comprises: [0217] calculating a time difference between the
reception time of the latest anchor frame and the reception time of
the first frame; [0218] estimating an expected number of frames
between the latest anchor frame and the first frame; and [0219]
estimating the offset value based on the time difference and an
expected time difference obtained based on the expected number of
frames.
[0220] EEE53. The method for controlling a jitter buffer according
to EEE 52, wherein the operation of estimating the expected number
of frames comprises obtaining the expected number of frames based
on sequence numbers of the latest anchor frame and the first frame
and information regarding silence frames between the previous
talkspurt and the current talkspurt, the information regarding
silence frames being carried in at least one frame in at least one
talkspurt in the previous talkspurt and the current talkspurt.
[0221] EEE54. The method for controlling a jitter buffer according
to EEE 53, wherein the information regarding silence frames
comprises the number of the silence frames, the number being
embedded in the first frame of the current talkspurt.
[0222] EEE55. The method for controlling a jitter buffer according
to EEE 53, wherein the information regarding silence frames
comprises timestamps in the last frame of the previous talkspurt
and the first frame in the current talkspurt.
[0223] EEE56. The method for controlling a jitter buffer according
to any one of EEE's 46-55, wherein the operation of estimating the
offset value further comprises adjusting the offset value of the
delay of the first frame in the current talkspurt based on an
average or weighted average of at least one earlier offset value,
or adopt the largest offset value among the present offset value
and the at least one earlier offset value.
[0224] EEE57. The method for controlling a jitter buffer according
to any one of EEE's 46-56, further comprising estimating the long
term length for each frame by calculating a statistic distribution
of history delay jitter values.
[0225] EEE58. The method for controlling a jitter buffer according
to any one of EEE's 46-57, further comprising adjusting the long
term length for each frame to a smaller value when the offset value
is greater than zero.
[0226] EEE59. The method for controlling a jitter buffer according
to EEE 58, further comprising fading the offset value with
time.
[0227] EEE60. The method for controlling a jitter buffer according
to any one of EEE's 57-59, wherein the operation of estimating the
long term length comprising resetting the long term length to an
initial value and resetting the history delay jitter values when
the long term length at the end of the last talkspurt is greater
than a predefined threshold.
[0228] EEE61. The method for controlling a jitter buffer according
to any one of EEE's 46-60, wherein the operation of adjusting the
length of the jitter buffer comprises setting the length of the
jitter buffer for the first frame of the current talkspurt based on
the long term length for the first frame and the number of frames
received at the same time with the first frame.
[0229] EEE62. The method for controlling a jitter buffer according
to any one of EEE's 46-61, further comprising: if the jitter buffer
length still used by the last talkspurt is greater than the long
term length and the first frame of the new talkspurt has arrived,
dropping some of the frames of the last talkspurt so that the
frames of the last talkspurt occupy only the long term length of
the jitter buffer.
[0230] EEE63. The method for controlling a jitter buffer according
to any one of EEE's 46-62, wherein the operation of adjusting the
length of the jitter buffer comprises adjusting the present length
of the jitter buffer toward the long term length.
[0231] EEE64. The method for controlling a jitter buffer according
to EEE 63, wherein the operation of adjusting the present length of
the jitter buffer toward the long term length comprises calculating
a weighted average of the long term length and the present
length.
[0232] EEE65. The method for controlling a jitter buffer according
to EEE 64, wherein the weight of the long term length is greater
when the long term length is greater than the present length, and
is smaller when the long term length is smaller than the present
length.
[0233] EEE66. The method for controlling a jitter buffer according
to any one of EEE's 46-65, wherein the operation of adjusting the
length of the jitter buffer comprises determining the length of the
jitter buffer for each frame as the long term length.
[0234] EEE67. A method for controlling a jitter buffer comprising:
[0235] estimating a long term length of the jitter buffer for each
frame by calculating a statistic distribution of history delay
jitter values; and [0236] setting the length of the jitter buffer
for the first frame of the current talkspurt based on the long term
length for the first frame and the number of frames received at the
same time with the first frame.
[0237] EEE68. The method for controlling a jitter buffer according
to EEE 67, further comprising: if the jitter buffer length still
used by the last talkspurt is greater than the long term length and
the first frame of the new talkspurt has arrived, dropping some of
the frames of the last talkspurt so that the frames of the last
talkspurt occupy only the long term length of the jitter
buffer.
[0238] EEE69. The method for controlling a jitter buffer according
to EEE 67 or 68, further comprising adjusting the present length of
the jitter buffer toward the long term length.
[0239] EEE70. The method for controlling a jitter buffer according
to EEE 69, wherein the operation of adjusting the present length of
the jitter buffer toward the long term length comprises calculating
a weighted average of the long term length and the present
length.
[0240] EEE71. The method for controlling a jitter buffer according
to EEE 70, wherein the weight of the long term length is greater
when the long term length is greater than the present length, and
is smaller when the long term length is smaller than the present
length.
[0241] EEE72. A method for controlling a jitter buffer comprising:
[0242] estimating a long term length of the jitter buffer for each
frame by calculating a statistic distribution of history delay
jitter values; and [0243] adjusting the present length of the
jitter buffer toward the long term length.
[0244] EEE73. The method for controlling a jitter buffer according
to EEE 72, wherein the operation of adjusting the present length of
the jitter buffer toward the long term length comprises calculating
a weighted average of the long term length and the present
length.
[0245] EEE74. The method for controlling a jitter buffer according
to EEE 73, wherein the weight of the long term length is greater
when the long term length is greater than the present length, and
is smaller when the long term length is smaller than the present
length.
[0246] EEE75. The method for controlling a jitter buffer according
to any one of EEe's 46-74, further comprising: [0247] monitoring
occupancy of the jitter buffer; and [0248] increasing the length of
the jitter buffer in response to the occupancy meeting a predefined
condition.
[0249] EEE76. The method for controlling a jitter buffer according
to EEE 75, wherein the operation of increasing the length of the
jitter buffer comprises increasing the length of the jitter buffer
where an occupancy less than a predetermined threshold has
consecutively occurred for a predetermined times.
[0250] EEE77. The method for controlling a jitter buffer according
to EEE 76, wherein the operation of increasing the length of the
jitter buffer comprises increasing the length of the jitter buffer
where a single-frame occupancy has consecutively occurred
twice.
[0251] EEE78. The method for controlling a jitter buffer according
to any one of EEE's 75-77, wherein the operation of increasing the
length of the jitter buffer is configured to increase the length of
the jitter buffer by one frame each time.
[0252] EEE79. The method for controlling a jitter buffer according
to any one of EEE's 75-77, further comprising monitoring the
content of the audio signal, and wherein the operation of
increasing the length of the jitter buffer is not performed where
the content of the audio signal is not appropriate for frame
insertion.
[0253] EEE80. The method for controlling a jitter buffer according
to EEE 79, wherein the operation of monitoring the content of the
audio signal comprises monitoring the transient state of the audio
signal, and wherein the operation of increasing the length of the
jitter buffer is not performed in transient signal area.
[0254] EEE81. The method for controlling a jitter buffer according
to any one of EEE's 75-77, wherein the operation of increasing the
length of the jitter buffer is not performed where the present
jitter buffer length exceeds a predetermined threshold.
[0255] EEE82. The method for controlling a jitter buffer according
to any one of EEE's 75-77, further comprising counting frame
insertions within the present talkspurt, and wherein the operation
of increasing the length of the jitter buffer is not performed
where the number of frame insertions within the present talkspurt
exceeds a predetermined threshold.
[0256] EEE83. A method for controlling a jitter buffer, comprising:
[0257] monitoring occupancy of the jitter buffer, and increasing
the length of the jitter buffer in response to the occupancy
meeting a predefined condition.
[0258] EEE84. The method for controlling a jitter buffer according
to EEE 83, wherein the operation of increasing the length of the
jitter buffer comprises increasing the length of the jitter buffer
where an occupancy less than a predetermined threshold has
consecutively occurred for a predetermined times.
[0259] EEE85. The method for controlling a jitter buffer according
to EEE 84, wherein the operation of increasing the length of the
jitter buffer comprises increasing the length of the jitter buffer
where a single-frame occupancy has consecutively occurred
twice.
[0260] EEE86. The method for controlling a jitter buffer according
to any one of EEE 83-85, wherein the operation of increasing the
length of the jitter buffer is configured to increase the length of
the jitter buffer by one frame each time.
[0261] EEE87. The method for controlling a jitter buffer according
to any one of EEE's 83-85, further comprising monitoring the
content of the audio signal, and wherein the operation of
increasing the length of the jitter buffer is not performed where
the content of the audio signal is not appropriate for frame
insertion.
[0262] EEE88. The method for controlling a jitter buffer according
to EEE 87, wherein the operation of monitoring the content of the
audio signal comprises monitoring the transient state of the audio
signal, and wherein the operation of increasing the length of the
jitter buffer is not performed in transient signal area.
[0263] EEE89. The method for controlling a jitter buffer according
to any one of EEE's 83-85, wherein the operation of increasing the
length of the jitter buffer is not performed where the present
jitter buffer length exceeds a predetermined threshold.
[0264] EEE90. The method for controlling a jitter buffer according
to any one of EEE's 83-85, further comprising counting frame
insertions within the present talkspurt, and wherein the operation
of increasing the length of the jitter buffer is not performed
where the number of frame insertions within the present talkspurt
exceeds a predetermined threshold.
[0265] EEE91. A computer-readable medium having computer program
instructions recorded thereon, when being executed by a processor,
the instructions enabling the processor to execute a method for
controlling a jitter buffer comprising: [0266] estimating an offset
value of the delay of a first frame in the current talkspurt with
respect to the delay of a latest anchor frame in a previous
talkspurt; and [0267] adjusting a length of the jitter buffer based
on a long term length of the jitter buffer for each frame and the
offset value.
[0268] EEE92. A computer-readable medium having computer program
instructions recorded thereon, when being executed by a processor,
the instructions enabling the processor to execute a method for
controlling a jitter buffer comprising: [0269] estimating a long
term length of the jitter buffer for each frame by calculating a
statistic distribution of history delay jitter values; and [0270]
setting the length of the jitter buffer for the first frame of the
current talkspurt based on the long term length for the first frame
and the number of frames received at the same time with the first
frame.
[0271] EEE93. A computer-readable medium having computer program
instructions recorded thereon, when being executed by a processor,
the instructions enabling the processor to execute a method for
controlling a jitter buffer comprising: [0272] estimating a long
term length of the jitter buffer for each frame by calculating a
statistic distribution of history delay jitter values; and [0273]
adjusting the present length of the jitter buffer toward the long
term length.
[0274] EEE94. A computer-readable medium having computer program
instructions recorded thereon, when being executed by a processor,
the instructions enabling the processor to execute a method for
controlling a jitter buffer, comprising: [0275] monitoring
occupancy of the jitter buffer, and [0276] increasing the length of
the jitter buffer in response to the occupancy meeting a predefined
condition.
[0277] Accordingly, the scope of the invention is to be construed
in accordance with the substance defined by the following
claims.
* * * * *