Request for Comments: 3551 Columbia University
Obsoletes: 1890 S. Casner
Category: Standards Track Packet Design
July 2003
RTP Profile for Audio and Video Conferences
with Minimal Control
Status of this Memo
This document specifies an Internet standards track protocol for the
Internet community, and requests discussion and suggestions for
improvements. Please refer to the current edition of the "Internet
Official Protocol Standards" (STD 1) for the standardization state
and status of this protocol. Distribution of this memo is unlimited.
Copyright Notice
Copyright (C) The Internet Society (2003). All Rights Reserved.
Abstract
This document describes a profile called "RTP/AVP" for the use of the
real-time transport protocol (RTP), version 2, and the associated
control protocol, RTCP, within audio and video multiparticipant
conferences with minimal control. It provides interpretations of
generic fields within the RTP specification suitable for audio and
video conferences. In particular, this document defines a set of
default mappings from payload type numbers to encodings.
This document also describes how audio and video data may be carried
within RTP. It defines a set of standard encodings and their names
when used within RTP. The descriptions provide pointers to reference
implementations and the detailed standards. This document is meant
as an aid for implementors of audio, video and other real-time
multimedia applications.
This memorandum obsoletes RFC1890. It is mostly backwards-
compatible except for functions removed because two interoperable
implementations were not found. The additions to RFC1890 codify
existing practice in the use of payload formats under this profile
and include new payload formats defined since RFC1890 was published.
Table of Contents
1. Introduction ................................................. 3
1.1 Terminology ............................................. 3
2. RTP and RTCP Packet Forms and Protocol Behavior .............. 4
3. Registering Additional Encodings ............................. 6
4. Audio ........................................................ 8
4.1 Encoding-Independent Rules .............................. 8
4.2 Operating Recommendations ............................... 9
4.3 Guidelines for Sample-Based Audio Encodings ............. 10
4.4 Guidelines for Frame-Based Audio Encodings .............. 11
4.5 Audio Encodings ......................................... 12
4.5.1 DVI4 ............................................ 13
4.5.2 G722 ............................................ 14
4.5.3 G723 ............................................ 14
4.5.4 G726-40, G726-32, G726-24, and G726-16 .......... 18
4.5.5 G728 ............................................ 19
4.5.6 G729 ............................................ 20
4.5.7 G729D and G729E ................................. 22
4.5.8 GSM ............................................. 24
4.5.9 GSM-EFR ......................................... 27
4.5.10 L8 .............................................. 27
4.5.11 L16 ............................................. 27
4.5.12 LPC ............................................. 27
4.5.13 MPA ............................................. 28
4.5.14 PCMA and PCMU ................................... 28
4.5.15 QCELP ........................................... 28
4.5.16 RED ............................................. 29
4.5.17 VDVI ............................................ 29
5. Video ........................................................ 30
5.1 CelB .................................................... 30
5.2 JPEG .................................................... 30
5.3 H261 .................................................... 30
5.4 H263 .................................................... 31
5.5 H263-1998 ............................................... 31
5.6 MPV ..................................................... 31
5.7 MP2T .................................................... 31
5.8 nv ...................................................... 32
6. Payload Type Definitions ..................................... 32
7. RTP over TCP and Similar Byte Stream Protocols ............... 34
8. Port Assignment .............................................. 34
9. Changes from RFC1890 ........................................ 35
10. Security Considerations ...................................... 38
11. IANA Considerations .......................................... 39
12. References ................................................... 39
12.1 Normative References .................................... 39
12.2 Informative References .................................. 39
13. Current Locations of Related Resources ....................... 41
14. Acknowledgments .............................................. 42
15. Intellectual Property Rights Statement ....................... 43
16. Authors' Addresses ........................................... 43
17. Full Copyright Statement ..................................... 44
1. Introduction
This profile defines aspects of RTP left unspecified in the RTP
Version 2 protocol definition (RFC3550) [1]. This profile is
intended for the use within audio and video conferences with minimal
session control. In particular, no support for the negotiation of
parameters or membership control is provided. The profile is
expected to be useful in sessions where no negotiation or membership
control are used (e.g., using the static payload types and the
membership indications provided by RTCP), but this profile may also
be useful in conjunction with a higher-level control protocol.
Use of this profile may be implicit in the use of the appropriate
applications; there may be no explicit indication by port number,
protocol identifier or the like. Applications such as session
directories may use the name for this profile specified in Section
11.
Other profiles may make different choices for the items specified
here.
This document also defines a set of encodings and payload formats for
audio and video. These payload format descriptions are included here
only as a matter of convenience since they are too small to warrant
separate documents. Use of these payload formats is NOT REQUIRED to
use this profile. Only the binding of some of the payload formats to
static payload type numbers in Tables 4 and 5 is normative.
1.1 Terminology
The key words "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT",
"SHOULD", "SHOULD NOT", "RECOMMENDED", "MAY", and "OPTIONAL" in this
document are to be interpreted as described in RFC2119 [2] and
indicate requirement levels for implementations compliant with this
RTP profile.
This document defines the term media type as dividing encodings of
audio and video content into three classes: audio, video and
audio/video (interleaved).
2. RTP and RTCP Packet Forms and Protocol Behavior
The section "RTP Profiles and Payload Format Specifications" of RFC
3550 enumerates a number of items that can be specified or modified
in a profile. This section addresses these items. Generally, this
profile follows the default and/or recommended aspects of the RTP
specification.
RTP data header: The standard format of the fixed RTP data
header is used (one marker bit).
Payload types: Static payload types are defined in Section 6.
RTP data header additions: No additional fixed fields are
appended to the RTP data header.
RTP data header extensions: No RTP header extensions are
defined, but applications operating under this profile MAY use
such extensions. Thus, applications SHOULD NOT assume that the
RTP header X bit is always zero and SHOULD be prepared to ignore
the header extension. If a header extension is defined in the
future, that definition MUST specify the contents of the first 16
bits in such a way that multiple different extensions can be
identified.
RTCP packet types: No additional RTCP packet types are defined
by this profile specification.
RTCP report interval: The suggested constants are to be used for
the RTCP report interval calculation. Sessions operating under
this profile MAY specify a separate parameter for the RTCP traffic
bandwidth rather than using the default fraction of the session
bandwidth. The RTCP traffic bandwidth MAY be divided into two
separate session parameters for those participants which are
active data senders and those which are not. Following the
recommendation in the RTP specification [1] that 1/4 of the RTCP
bandwidth be dedicated to data senders, the RECOMMENDED default
values for these two parameters would be 1.25% and 3.75%,
respectively. For a particular session, the RTCP bandwidth for
non-data-senders MAY be set to zero when operating on
unidirectional links or for sessions that don't require feedback
on the quality of reception. The RTCP bandwidth for data senders
SHOULD be kept non-zero so that sender reports can still be sent
for inter-media synchronization and to identify the source by
CNAME. The means by which the one or two session parameters for
RTCP bandwidth are specified is beyond the scope of this memo.
SR/RR extension: No extension section is defined for the RTCP SR
or RR packet.
SDES use: Applications MAY use any of the SDES items described
in the RTP specification. While CNAME information MUST be sent
every reporting interval, other items SHOULD only be sent every
third reporting interval, with NAME sent seven out of eight times
within that slot and the remaining SDES items cyclically taking up
the eighth slot, as defined in Section 6.2.2 of the RTP
specification. In other words, NAME is sent in RTCP packets 1, 4,
7, 10, 13, 16, 19, while, say, EMAIL is used in RTCP packet 22.
Security: The RTP default security services are also the default
under this profile.
String-to-key mapping: No mapping is specified by this profile.
Congestion: RTP and this profile may be used in the context of
enhanced network service, for example, through Integrated Services
(RFC1633) [4] or Differentiated Services (RFC2475) [5], or they
may be used with best effort service.
If enhanced service is being used, RTP receivers SHOULD monitor
packet loss to ensure that the service that was requested is
actually being delivered. If it is not, then they SHOULD assume
that they are receiving best-effort service and behave
accordingly.
If best-effort service is being used, RTP receivers SHOULD monitor
packet loss to ensure that the packet loss rate is within
acceptable parameters. Packet loss is considered acceptable if a
TCP flow across the same network path and experiencing the same
network conditions would achieve an average throughput, measured
on a reasonable timescale, that is not less than the RTP flow is
achieving. This condition can be satisfied by implementing
congestion control mechanisms to adapt the transmission rate (or
the number of layers subscribed for a layered multicast session),
or by arranging for a receiver to leave the session if the loss
rate is unacceptably high.
The comparison to TCP cannot be specified exactly, but is intended
as an "order-of-magnitude" comparison in timescale and throughput.
The timescale on which TCP throughput is measured is the round-
trip time of the connection. In essence, this requirement states
that it is not acceptable to deploy an application (using RTP or
any other transport protocol) on the best-effort Internet which
consumes bandwidth arbitrarily and does not compete fairly with
TCP within an order of magnitude.
Underlying protocol: The profile specifies the use of RTP over
unicast and multicast UDP as well as TCP. (This does not preclude
the use of these definitions when RTP is carried by other lower-
layer protocols.)
Transport mapping: The standard mapping of RTP and RTCP to
transport-level addresses is used.
Encapsulation: This profile leaves to applications the
specification of RTP encapsulation in protocols other than UDP.
3. Registering Additional Encodings
This profile lists a set of encodings, each of which is comprised of
a particular media data compression or representation plus a payload
format for encapsulation within RTP. Some of those payload formats
are specified here, while others are specified in separate RFCs. It
is expected that additional encodings beyond the set listed here will
be created in the future and specified in additional payload format
RFCs.
This profile also assigns to each encoding a short name which MAY be
used by higher-level control protocols, such as the Session
Description Protocol (SDP), RFC2327 [6], to identify encodings
selected for a particular RTP session.
In some contexts it may be useful to refer to these encodings in the
form of a MIME content-type. To facilitate this, RFC3555 [7]
provides registrations for all of the encodings names listed here as
MIME subtype names under the "audio" and "video" MIME types through
the MIME registration procedure as specified in RFC2048 [8].
Any additional encodings specified for use under this profile (or
others) may also be assigned names registered as MIME subtypes with
the Internet Assigned Numbers Authority (IANA). This registry
provides a means to insure that the names assigned to the additional
encodings are kept unique. RFC3555 specifies the information that
is required for the registration of RTP encodings.
In addition to assigning names to encodings, this profile also
assigns static RTP payload type numbers to some of them. However,
the payload type number space is relatively small and cannot
accommodate assignments for all existing and future encodings.
During the early stages of RTP development, it was necessary to use
statically assigned payload types because no other mechanism had been
specified to bind encodings to payload types. It was anticipated
that non-RTP means beyond the scope of this memo (such as directory
services or invitation protocols) would be specified to establish a
dynamic mapping between a payload type and an encoding. Now,
mechanisms for defining dynamic payload type bindings have been
specified in the Session Description Protocol (SDP) and in other
protocols such as ITU-T Recommendation H.323/H.245. These mechanisms
associate the registered name of the encoding/payload format, along
with any additional required parameters, such as the RTP timestamp
clock rate and number of channels, with a payload type number. This
association is effective only for the duration of the RTP session in
which the dynamic payload type binding is made. This association
applies only to the RTP session for which it is made, thus the
numbers can be re-used for different encodings in different sessions
so the number space limitation is avoided.
This profile reserves payload type numbers in the range 96-127
exclusively for dynamic assignment. Applications SHOULD first use
values in this range for dynamic payload types. Those applications
which need to define more than 32 dynamic payload types MAY bind
codes below 96, in which case it is RECOMMENDED that unassigned
payload type numbers be used first. However, the statically assigned
payload types are default bindings and MAY be dynamically bound to
new encodings if needed. Redefining payload types below 96 may cause
incorrect operation if an attempt is made to join a session without
obtaining session description information that defines the dynamic
payload types.
Dynamic payload types SHOULD NOT be used without a well-defined
mechanism to indicate the mapping. Systems that expect to
interoperate with others operating under this profile SHOULD NOT make
their own assignments of proprietary encodings to particular, fixed
payload types.
This specification establishes the policy that no additional static
payload types will be assigned beyond the ones defined in this
document. Establishing this policy avoids the problem of trying to
create a set of criteria for accepting static assignments and
encourages the implementation and deployment of the dynamic payload
type mechanisms.
The final set of static payload type assignments is provided in
Tables 4 and 5.
4. Audio
4.1 Encoding-Independent Rules
Since the ability to suppress silence is one of the primary
motivations for using packets to transmit voice, the RTP header
carries both a sequence number and a timestamp to allow a receiver to
distinguish between lost packets and periods of time when no data was
transmitted. Discontiguous transmission (silence suppression) MAY be
used with any audio payload format. Receivers MUST assume that
senders may suppress silence unless this is restricted by signaling
specified elsewhere. (Even if the transmitter does not suppress
silence, the receiver should be prepared to handle periods when no
data is present since packets may be lost.)
Some payload formats (see Sections 4.5.3 and 4.5.6) define a "silence
insertion descriptor" or "comfort noise" frame to specify parameters
for artificial noise that may be generated during a period of silence
to approximate the background noise at the source. For other payload
formats, a generic Comfort Noise (CN) payload format is specified in
RFC3389 [9]. When the CN payload format is used with another
payload format, different values in the RTP payload type field
distinguish comfort-noise packets from those of the selected payload
format.
For applications which send either no packets or occasional comfort-
noise packets during silence, the first packet of a talkspurt, that
is, the first packet after a silence period during which packets have
not been transmitted contiguously, SHOULD be distinguished by setting
the marker bit in the RTP data header to one. The marker bit in all
other packets is zero. The beginning of a talkspurt MAY be used to
adjust the playout delay to reflect changing network delays.
Applications without silence suppression MUST set the marker bit to
zero.
The RTP clock rate used for generating the RTP timestamp is
independent of the number of channels and the encoding; it usually
equals the number of sampling periods per second. For N-channel
encodings, each sampling period (say, 1/8,000 of a second) generates
N samples. (This terminology is standard, but somewhat confusing, as
the total number of samples generated per second is then the sampling
rate times the channel count.)
If multiple audio channels are used, channels are numbered left-to-
right, starting at one. In RTP audio packets, information from
lower-numbered channels precedes that from higher-numbered channels.
For more than two channels, the convention followed by the AIFF-C
audio interchange format SHOULD be followed [3], using the following
notation, unless some other convention is specified for a particular
encoding or payload format:
l left
r right
c center
S surround
F front
R rear
channels description channel
1 2 3 4 5 6
_________________________________________________
2 stereo l r
3 l r c
4 l c r S
5 Fl Fr Fc Sl Sr
6 l lc c r rc S
Note: RFC1890 defined two conventions for the ordering of four
audio channels. Since the ordering is indicated implicitly by
the number of channels, this was ambiguous. In this revision,
the order described as "quadrophonic" has been eliminated to
remove the ambiguity. This choice was based on the observation
that quadrophonic consumer audio format did not become popular
whereas surround-sound subsequently has.
Samples for all channels belonging to a single sampling instant MUST
be within the same packet. The interleaving of samples from
different channels depends on the encoding. General guidelines are
given in Section 4.3 and 4.4.
The sampling frequency SHOULD be drawn from the set: 8,000, 11,025,
16,000, 22,050, 24,000, 32,000, 44,100 and 48,000 Hz. (Older Apple
Macintosh computers had a native sample rate of 22,254.54 Hz, which
can be converted to 22,050 with acceptable quality by dropping 4
samples in a 20 ms frame.) However, most audio encodings are defined
for a more restricted set of sampling frequencies. Receivers SHOULD
be prepared to accept multi-channel audio, but MAY choose to only
play a single channel.
4.2 Operating Recommendations
The following recommendations are default operating parameters.
Applications SHOULD be prepared to handle other values. The ranges
given are meant to give guidance to application writers, allowing a
set of applications conforming to these guidelines to interoperate
without additional negotiation. These guidelines are not intended to
restrict operating parameters for applications that can negotiate a
set of interoperable parameters, e.g., through a conference control
protocol.
For packetized audio, the default packetization interval SHOULD have
a duration of 20 ms or one frame, whichever is longer, unless
otherwise noted in Table 1 (column "ms/packet"). The packetization
interval determines the minimum end-to-end delay; longer packets
introduce less header overhead but higher delay and make packet loss
more noticeable. For non-interactive applications such as lectures
or for links with severe bandwidth constraints, a higher
packetization delay MAY be used. A receiver SHOULD accept packets
representing between 0 and 200 ms of audio data. (For framed audio
encodings, a receiver SHOULD accept packets with a number of frames
equal to 200 ms divided by the frame duration, rounded up.) This
restriction allows reasonable buffer sizing for the receiver.
4.3 Guidelines for Sample-Based Audio Encodings
In sample-based encodings, each audio sample is represented by a
fixed number of bits. Within the compressed audio data, codes for
individual samples may span octet boundaries. An RTP audio packet
may contain any number of audio samples, subject to the constraint
that the number of bits per sample times the number of samples per
packet yields an integral octet count. Fractional encodings produce
less than one octet per sample.
The duration of an audio packet is determined by the number of
samples in the packet.
For sample-based encodings producing one or more octets per sample,
samples from different channels sampled at the same sampling instant
SHOULD be packed in consecutive octets. For example, for a two-
channel encoding, the octet sequence is (left channel, first sample),
(right channel, first sample), (left channel, second sample), (right
channel, second sample), .... For multi-octet encodings, octets
SHOULD be transmitted in network byte order (i.e., most significant
octet first).
The packing of sample-based encodings producing less than one octet
per sample is encoding-specific.
The RTP timestamp reflects the instant at which the first sample in
the packet was sampled, that is, the oldest information in the
packet.
4.4 Guidelines for Frame-Based Audio Encodings
Frame-based encodings encode a fixed-length block of audio into
another block of compressed data, typically also of fixed length.
For frame-based encodings, the sender MAY choose to combine several
such frames into a single RTP packet. The receiver can tell the
number of frames contained in an RTP packet, if all the frames have
the same length, by dividing the RTP payload length by the audio
frame size which is defined as part of the encoding. This does not
work when carrying frames of different sizes unless the frame sizes
are relatively prime. If not, the frames MUST indicate their size.
For frame-based codecs, the channel order is defined for the whole
block. That is, for two-channel audio, right and left samples SHOULD
be coded independently, with the encoded frame for the left channel
preceding that for the right channel.
All frame-oriented audio codecs SHOULD be able to encode and decode
several consecutive frames within a single packet. Since the frame
size for the frame-oriented codecs is given, there is no need to use
a separate designation for the same encoding, but with different
number of frames per packet.
RTP packets SHALL contain a whole number of frames, with frames
inserted according to age within a packet, so that the oldest frame
(to be played first) occurs immediately after the RTP packet header.
The RTP timestamp reflects the instant at which the first sample in
the first frame was sampled, that is, the oldest information in the
packet.
4.5 Audio Encodings
name of sampling default
encoding sample/frame bits/sample rate ms/frame ms/packet
__________________________________________________________________
DVI4 sample 4 var. 20
G722 sample 8 16,000 20
G723 frame N/A 8,000 30 30
G726-40 sample 5 8,000 20
G726-32 sample 4 8,000 20
G726-24 sample 3 8,000 20
G726-16 sample 2 8,000 20
G728 frame N/A 8,000 2.5 20
G729 frame N/A 8,000 10 20
G729D frame N/A 8,000 10 20
G729E frame N/A 8,000 10 20
GSM frame N/A 8,000 20 20
GSM-EFR frame N/A 8,000 20 20
L8 sample 8 var. 20
L16 sample 16 var. 20
LPC frame N/A 8,000 20 20
MPA frame N/A var. var.
PCMA sample 8 var. 20
PCMU sample 8 var. 20
QCELP frame N/A 8,000 20 20
VDVI sample var. var. 20
Table 1: Properties of Audio Encodings (N/A: not applicable; var.:
variable)
The characteristics of the audio encodings described in this document
are shown in Table 1; they are listed in order of their payload type
in Table 4. While most audio codecs are only specified for a fixed
sampling rate, some sample-based algorithms (indicated by an entry of
"var." in the sampling rate column of Table 1) may be used with
different sampling rates, resulting in different coded bit rates.
When used with a sampling rate other than that for which a static
payload type is defined, non-RTP means beyond the scope of this memo
MUST be used to define a dynamic payload type and MUST indicate the
selected RTP timestamp clock rate, which is usually the same as the
sampling rate for audio.
4.5.1 DVI4
DVI4 uses an adaptive delta pulse code modulation (ADPCM) encoding
scheme that was specified by the Interactive Multimedia Association
(IMA) as the "IMA ADPCM wave type". However, the encoding defined
here as DVI4 differs in three respects from the IMA specification:
o The RTP DVI4 header contains the predicted value rather than the
first sample value contained the IMA ADPCM block header.
o IMA ADPCM blocks contain an odd number of samples, since the first
sample of a block is contained just in the header (uncompressed),
followed by an even number of compressed samples. DVI4 has an
even number of compressed samples only, using the `predict' word
from the header to decode the first sample.
o For DVI4, the 4-bit samples are packed with the first sample in
the four most significant bits and the second sample in the four
least significant bits. In the IMA ADPCM codec, the samples are
packed in the opposite order.
Each packet contains a single DVI block. This profile only defines
the 4-bit-per-sample version, while IMA also specified a 3-bit-per-
sample encoding.
The "header" word for each channel has the following structure:
int16 predict; /* predicted value of first sample
from the previous block (L16 format) */
u_int8 index; /* current index into stepsize table */
u_int8 reserved; /* set to zero by sender, ignored by receiver */
Each octet following the header contains two 4-bit samples, thus the
number of samples per packet MUST be even because there is no means
to indicate a partially filled last octet.
Packing of samples for multiple channels is for further study.
The IMA ADPCM algorithm was described in the document IMA Recommended
Practices for Enhancing Digital Audio Compatibility in Multimedia
Systems (version 3.0). However, the Interactive Multimedia
Association ceased operations in 1997. Resources for an archived
copy of that document and a software implementation of the RTP DVI4
encoding are listed in Section 13.
4.5.2 G722
G722 is specified in ITU-T Recommendation G.722, "7 kHz audio-coding
within 64 kbit/s". The G.722 encoder produces a stream of octets,
each of which SHALL be octet-aligned in an RTP packet. The first bit
transmitted in the G.722 octet, which is the most significant bit of
the higher sub-band sample, SHALL correspond to the most significant
bit of the octet in the RTP packet.
Even though the actual sampling rate for G.722 audio is 16,000 Hz,
the RTP clock rate for the G722 payload format is 8,000 Hz because
that value was erroneously assigned in RFC1890 and must remain
unchanged for backward compatibility. The octet rate or sample-pair
rate is 8,000 Hz.
4.5.3 G723
G723 is specified in ITU Recommendation G.723.1, "Dual-rate speech
coder for multimedia communications transmitting at 5.3 and 6.3
kbit/s". The G.723.1 5.3/6.3 kbit/s codec was defined by the ITU-T
as a mandatory codec for ITU-T H.324 GSTN videophone terminal
applications. The algorithm has a floating point specification in
Annex B to G.723.1, a silence compression algorithm in Annex A to
G.723.1 and a scalable channel coding scheme for wireless
applications in G.723.1 Annex C.
This Recommendation specifies a coded representation that can be used
for compressing the speech signal component of multi-media services
at a very low bit rate. Audio is encoded in 30 ms frames, with an
additional delay of 7.5 ms due to look-ahead. A G.723.1 frame can be
one of three sizes: 24 octets (6.3 kb/s frame), 20 octets (5.3 kb/s
frame), or 4 octets. These 4-octet frames are called SID frames
(Silence Insertion Descriptor) and are used to specify comfort noise
parameters. There is no restriction on how 4, 20, and 24 octet
frames are intermixed. The least significant two bits of the first
octet in the frame determine the frame size and codec type:
bits content octets/frame
00 high-rate speech (6.3 kb/s) 24
01 low-rate speech (5.3 kb/s) 20
10 SID frame 4
11 reserved
It is possible to switch between the two rates at any 30 ms frame
boundary. Both (5.3 kb/s and 6.3 kb/s) rates are a mandatory part of
the encoder and decoder. Receivers MUST accept both data rates and
MUST accept SID frames unless restriction of these capabilities has
been signaled. The MIME registration for G723 in RFC3555 [7]
specifies parameters that MAY be used with MIME or SDP to restrict to
a single data rate or to restrict the use of SID frames. This coder
was optimized to represent speech with near-toll quality at the above
rates using a limited amount of complexity.
The packing of the encoded bit stream into octets and the
transmission order of the octets is specified in Rec. G.723.1 and is
the same as that produced by the G.723 C code reference
implementation. For the 6.3 kb/s data rate, this packing is
illustrated as follows, where the header (HDR) bits are always "0 0"
as shown in Fig. 1 to indicate operation at 6.3 kb/s, and the Z bit
is always set to zero. The diagrams show the bit packing in "network
byte order", also known as big-endian order. The bits of each 32-bit
word are numbered 0 to 31, with the most significant bit on the left
and numbered 0. The octets (bytes) of each word are transmitted most
significant octet first. The bits of each data field are numbered in
the order of the bit stream representation of the encoding (least
significant bit first). The vertical bars indicate the boundaries
between field fragments.
0 1 2 3
0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
| LPC |HDR| LPC | LPC | ACL0 |LPC|
| | | | | | |
|0 0 0 0 0 0|0 0|1 1 1 1 0 0 0 0|2 2 1 1 1 1 1 1|0 0 0 0 0 0|2 2|
|5 4 3 2 1 0| |3 2 1 0 9 8 7 6|1 0 9 8 7 6 5 4|5 4 3 2 1 0|3 2|
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
| ACL2 |ACL|A| GAIN0 |ACL|ACL| GAIN0 | GAIN1 |
| | 1 |C| | 3 | 2 | | |
|0 0 0 0 0|0 0|0|0 0 0 0|0 0|0 0|1 1 0 0 0 0 0 0|0 0 0 0 0 0 0 0|
|4 3 2 1 0|1 0|6|3 2 1 0|1 0|6 5|1 0 9 8 7 6 5 4|7 6 5 4 3 2 1 0|
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
| GAIN2 | GAIN1 | GAIN2 | GAIN3 | GRID | GAIN3 |
| | | | | | |
|0 0 0 0|1 1 0 0|1 1 0 0 0 0 0 0|0 0 0 0 0 0 0 0|0 0 0 0|1 1 0 0|
|3 2 1 0|1 0 9 8|1 0 9 8 7 6 5 4|7 6 5 4 3 2 1 0|3 2 1 0|1 0 9 8|
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
| MSBPOS |Z|POS| MSBPOS | POS0 |POS| POS0 |
| | | 0 | | | 1 | |
|0 0 0 0 0 0 0|0|0 0|1 1 1 0 0 0|0 0 0 0 0 0 0 0|0 0|1 1 1 1 1 1|
|6 5 4 3 2 1 0| |1 0|2 1 0 9 8 7|9 8 7 6 5 4 3 2|1 0|5 4 3 2 1 0|
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
| POS1 | POS2 | POS1 | POS2 | POS3 | POS2 |
| | | | | | |
|0 0 0 0 0 0 0 0|0 0 0 0|1 1 1 1|1 1 0 0 0 0 0 0|0 0 0 0|1 1 1 1|
|9 8 7 6 5 4 3 2|3 2 1 0|3 2 1 0|1 0 9 8 7 6 5 4|3 2 1 0|5 4 3 2|
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
| POS3 | PSIG0 |POS|PSIG2| PSIG1 | PSIG3 |PSIG2|
| | | 3 | | | | |
|1 1 0 0 0 0 0 0|0 0 0 0 0 0|1 1|0 0 0|0 0 0 0 0|0 0 0 0 0|0 0 0|
|1 0 9 8 7 6 5 4|5 4 3 2 1 0|3 2|2 1 0|4 3 2 1 0|4 3 2 1 0|5 4 3|
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
Figure 1: G.723 (6.3 kb/s) bit packing
For the 5.3 kb/s data rate, the header (HDR) bits are always "0 1",
as shown in Fig. 2, to indicate operation at 5.3 kb/s.
0 1 2 3
0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
| LPC |HDR| LPC | LPC | ACL0 |LPC|
| | | | | | |
|0 0 0 0 0 0|0 1|1 1 1 1 0 0 0 0|2 2 1 1 1 1 1 1|0 0 0 0 0 0|2 2|
|5 4 3 2 1 0| |3 2 1 0 9 8 7 6|1 0 9 8 7 6 5 4|5 4 3 2 1 0|3 2|
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
| ACL2 |ACL|A| GAIN0 |ACL|ACL| GAIN0 | GAIN1 |
| | 1 |C| | 3 | 2 | | |
|0 0 0 0 0|0 0|0|0 0 0 0|0 0|0 0|1 1 0 0 0 0 0 0|0 0 0 0 0 0 0 0|
|4 3 2 1 0|1 0|6|3 2 1 0|1 0|6 5|1 0 9 8 7 6 5 4|7 6 5 4 3 2 1 0|
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
| GAIN2 | GAIN1 | GAIN2 | GAIN3 | GRID | GAIN3 |
| | | | | | |
|0 0 0 0|1 1 0 0|1 1 0 0 0 0 0 0|0 0 0 0 0 0 0 0|0 0 0 0|1 1 0 0|
|3 2 1 0|1 0 9 8|1 0 9 8 7 6 5 4|7 6 5 4 3 2 1 0|4 3 2 1|1 0 9 8|
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
| POS0 | POS1 | POS0 | POS1 | POS2 |
| | | | | |
|0 0 0 0 0 0 0 0|0 0 0 0|1 1 0 0|1 1 0 0 0 0 0 0|0 0 0 0 0 0 0 0|
|7 6 5 4 3 2 1 0|3 2 1 0|1 0 9 8|1 0 9 8 7 6 5 4|7 6 5 4 3 2 1 0|
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
| POS3 | POS2 | POS3 | PSIG1 | PSIG0 | PSIG3 | PSIG2 |
| | | | | | | |
|0 0 0 0|1 1 0 0|1 1 0 0 0 0 0 0|0 0 0 0|0 0 0 0|0 0 0 0|0 0 0 0|
|3 2 1 0|1 0 9 8|1 0 9 8 7 6 5 4|3 2 1 0|3 2 1 0|3 2 1 0|3 2 1 0|
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
Figure 2: G.723 (5.3 kb/s) bit packing
The packing of G.723.1 SID (silence) frames, which are indicated by
the header (HDR) bits having the pattern "1 0", is depicted in Fig.
3.
0 1 2 3
0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
| LPC |HDR| LPC | LPC | GAIN |LPC|
| | | | | | |
|0 0 0 0 0 0|1 0|1 1 1 1 0 0 0 0|2 2 1 1 1 1 1 1|0 0 0 0 0 0|2 2|
|5 4 3 2 1 0| |3 2 1 0 9 8 7 6|1 0 9 8 7 6 5 4|5 4 3 2 1 0|3 2|
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
Figure 3: G.723 SID mode bit packing
4.5.4 G726-40, G726-32, G726-24, and G726-16
ITU-T Recommendation G.726 describes, among others, the algorithm
recommended for conversion of a single 64 kbit/s A-law or mu-law PCM
channel encoded at 8,000 samples/sec to and from a 40, 32, 24, or 16
kbit/s channel. The conversion is applied to the PCM stream using an
Adaptive Differential Pulse Code Modulation (ADPCM) transcoding
technique. The ADPCM representation consists of a series of
codewords with a one-to-one correspondence to the samples in the PCM
stream. The G726 data rates of 40, 32, 24, and 16 kbit/s have
codewords of 5, 4, 3, and 2 bits, respectively.
The 16 and 24 kbit/s encodings do not provide toll quality speech.
They are designed for used in overloaded Digital Circuit
Multiplication Equipment (DCME). ITU-T G.726 recommends that the 16
and 24 kbit/s encodings should be alternated with higher data rate
encodings to provide an average sample size of between 3.5 and 3.7
bits per sample.
The encodings of G.726 are here denoted as G726-40, G726-32, G726-24,
and G726-16. Prior to 1990, G721 described the 32 kbit/s ADPCM
encoding, and G723 described the 40, 32, and 16 kbit/s encodings.
Thus, G726-32 designates the same algorithm as G721 in RFC1890.
A stream of G726 codewords contains no information on the encoding
being used, therefore transitions between G726 encoding types are not
permitted within a sequence of packed codewords. Applications MUST
determine the encoding type of packed codewords from the RTP payload
identifier.
No payload-specific header information SHALL be included as part of
the audio data. A stream of G726 codewords MUST be packed into
octets as follows: the first codeword is placed into the first octet
such that the least significant bit of the codeword aligns with the
least significant bit in the octet, the second codeword is then
packed so that its least significant bit coincides with the least
significant unoccupied bit in the octet. When a complete codeword
cannot be placed into an octet, the bits overlapping the octet
boundary are placed into the least significant bits of the next
octet. Packing MUST end with a completely packed final octet. The
number of codewords packed will therefore be a multiple of 8, 2, 8,
and 4 for G726-40, G726-32, G726-24, and G726-16, respectively. An
example of the packing scheme for G726-32 codewords is as shown,
where bit 7 is the least significant bit of the first octet, and bit
A3 is the least significant bit of the first codeword:
0 1
0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-
|B B B B|A A A A|D D D D|C C C C| ...
|0 1 2 3|0 1 2 3|0 1 2 3|0 1 2 3|
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-
An example of the packing scheme for G726-24 codewords follows, where
again bit 7 is the least significant bit of the first octet, and bit
A2 is the least significant bit of the first codeword:
0 1 2
0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-
|C C|B B B|A A A|F|E E E|D D D|C|H H H|G G G|F F| ...
|1 2|0 1 2|0 1 2|2|0 1 2|0 1 2|0|0 1 2|0 1 2|0 1|
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-
Note that the "little-endian" direction in which samples are packed
into octets in the G726-16, -24, -32 and -40 payload formats
specified here is consistent with ITU-T Recommendation X.420, but is
the opposite of what is specified in ITU-T Recommendation I.366.2
Annex E for ATM AAL2 transport. A second set of RTP payload formats
matching the packetization of I.366.2 Annex E and identified by MIME
subtypes AAL2-G726-16, -24, -32 and -40 will be specified in a
separate document.
4.5.5 G728
G728 is specified in ITU-T Recommendation G.728, "Coding of speech at
16 kbit/s using low-delay code excited linear prediction".
A G.278 encoder translates 5 consecutive audio samples into a 10-bit
codebook index, resulting in a bit rate of 16 kb/s for audio sampled
at 8,000 samples per second. The group of five consecutive samples
is called a vector. Four consecutive vectors, labeled V1 to V4
(where V1 is to be played first by the receiver), build one G.728
frame. The four vectors of 40 bits are packed into 5 octets, labeled
B1 through B5. B1 SHALL be placed first in the RTP packet.
Referring to the figure below, the principle for bit order is
"maintenance of bit significance". Bits from an older vector are
more significant than bits from newer vectors. The MSB of the frame
goes to the MSB of B1 and the LSB of the frame goes to LSB of B5.
1 2 3 3
0 0 0 0 9
++++++++++++++++++++++++++++++++++++++++
<---V1---><---V2---><---V3---><---V4---> vectors
<--B1--><--B2--><--B3--><--B4--><--B5--> octets
<------------- frame 1 ---------------->
In particular, B1 contains the eight most significant bits of V1,
with the MSB of V1 being the MSB of B1. B2 contains the two least
significant bits of V1, the more significant of the two in its MSB,
and the six most significant bits of V2. B1 SHALL be placed first in
the RTP packet and B5 last.
4.5.6 G729
G729 is specified in ITU-T Recommendation G.729, "Coding of speech at
8 kbit/s using conjugate structure-algebraic code excited linear
prediction (CS-ACELP)". A reduced-complexity version of the G.729
algorithm is specified in Annex A to Rec. G.729. The speech coding
algorithms in the main body of G.729 and in G.729 Annex A are fully
interoperable with each other, so there is no need to further
distinguish between them. An implementation that signals or accepts
use of G729 payload format may implement either G.729 or G.729A
unless restricted by additional signaling specified elsewhere related
specifically to the encoding rather than the payload format. The
G.729 and G.729 Annex A codecs were optimized to represent speech
with high quality, where G.729 Annex A trades some speech quality for
an approximate 50% complexity reduction [10]. See the next Section
(4.5.7) for other data rates added in later G.729 Annexes. For all
data rates, the sampling frequency (and RTP timestamp clock rate) is
8,000 Hz.
A voice activity detector (VAD) and comfort noise generator (CNG)
algorithm in Annex B of G.729 is RECOMMENDED for digital simultaneous
voice and data applications and can be used in conjunction with G.729
or G.729 Annex A. A G.729 or G.729 Annex A frame contains 10 octets,
while the G.729 Annex B comfort noise frame occupies 2 octets.
Receivers MUST accept comfort noise frames if restriction of their
use has not been signaled. The MIME registration for G729 in RFC
3555 [7] specifies a parameter that MAY be used with MIME or SDP to
restrict the use of comfort noise frames.
A G729 RTP packet may consist of zero or more G.729 or G.729 Annex A
frames, followed by zero or one G.729 Annex B frames. The presence
of a comfort noise frame can be deduced from the length of the RTP
payload. The default packetization interval is 20 ms (two frames),
but in some situations it may be desirable to send 10 ms packets. An
example would be a transition from speech to comfort noise in the
first 10 ms of the packet. For some applications, a longer
packetization interval may be required to reduce the packet rate.
0 1 2 3
0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
|L| L1 | L2 | L3 | P1 |P| C1 |
|0| | | | |0| |
| |0 1 2 3 4 5 6|0 1 2 3 4|0 1 2 3 4|0 1 2 3 4 5 6 7| |0 1 2 3 4|
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
| C1 | S1 | GA1 | GB1 | P2 | C2 |
| 1 1 1| | | | | |
|5 6 7 8 9 0 1 2|0 1 2 3|0 1 2|0 1 2 3|0 1 2 3 4|0 1 2 3 4 5 6 7|
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
| C2 | S2 | GA2 | GB2 |
| 1 1 1| | | |
|8 9 0 1 2|0 1 2 3|0 1 2|0 1 2 3|
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
Figure 4: G.729 and G.729A bit packing
The transmitted parameters of a G.729/G.729A 10-ms frame, consisting
of 80 bits, are defined in Recommendation G.729, Table 8/G.729. The
mapping of the these parameters is given below in Fig. 4. The
diagrams show the bit packing in "network byte order", also known as
big-endian order. The bits of each 32-bit word are numbered 0 to 31,
with the most significant bit on the left and numbered 0. The octets
(bytes) of each word are transmitted most significant octet first.
The bits of each data field are numbered in the order as produced by
the G.729 C code reference implementation.
The packing of the G.729 Annex B comfort noise frame is shown in Fig.
5.
0 1
0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
|L| LSF1 | LSF2 | GAIN |R|
|S| | | |E|
|F| | | |S|
|0|0 1 2 3 4|0 1 2 3|0 1 2 3 4|V| RESV = Reserved (zero)
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
Figure 5: G.729 Annex B bit packing
4.5.7 G729D and G729E
Annexes D and E to ITU-T Recommendation G.729 provide additional data
rates. Because the data rate is not signaled in the bitstream, the
different data rates are given distinct RTP encoding names which are
mapped to distinct payload type numbers. G729D indicates a 6.4
kbit/s coding mode (G.729 Annex D, for momentary reduction in channel
capacity), while G729E indicates an 11.8 kbit/s mode (G.729 Annex E,
for improved performance with a wide range of narrow-band input
signals, e.g., music and background noise). Annex E has two
operating modes, backward adaptive and forward adaptive, which are
signaled by the first two bits in each frame (the most significant
two bits of the first octet).
The voice activity detector (VAD) and comfort noise generator (CNG)
algorithm specified in Annex B of G.729 may be used with Annex D and
Annex E frames in addition to G.729 and G.729 Annex A frames. The
algorithm details for the operation of Annexes D and E with the Annex
B CNG are specified in G.729 Annexes F and G. Note that Annexes F
and G do not introduce any new encodings. Receivers MUST accept
comfort noise frames if restriction of their use has not been
signaled. The MIME registrations for G729D and G729E in RFC3555 [7]
specify a parameter that MAY be used with MIME or SDP to restrict the
use of comfort noise frames.
For G729D, an RTP packet may consist of zero or more G.729 Annex D
frames, followed by zero or one G.729 Annex B frame. Similarly, for
G729E, an RTP packet may consist of zero or more G.729 Annex E
frames, followed by zero or one G.729 Annex B frame. The presence of
a comfort noise frame can be deduced from the length of the RTP
payload.
A single RTP packet must contain frames of only one data rate,
optionally followed by one comfort noise frame. The data rate may be
changed from packet to packet by changing the payload type number.
G.729 Annexes D, E and H describe what the encoding and decoding
algorithms must do to accommodate a change in data rate.
For G729D, the bits of a G.729 Annex D frame are formatted as shown
below in Fig. 6 (cf. Table D.1/G.729). The frame length is 64 bits.
0 1 2 3
0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
|L| L1 | L2 | L3 | P1 | C1 |
|0| | | | | |
| |0 1 2 3 4 5 6|0 1 2 3 4|0 1 2 3 4|0 1 2 3 4 5 6 7|0 1 2 3 4 5|
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
| C1 |S1 | GA1 | GB1 | P2 | C2 |S2 | GA2 | GB2 |
| | | | | | | | | |
|6 7 8|0 1|0 1 2|0 1 2|0 1 2 3|0 1 2 3 4 5 6 7 8|0 1|0 1 2|0 1 2|
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+