Internet Draft                                               Adam H. Li
draft-ietf-avt-evrc-smv-00.txt                                     UCLA
February 4, 2002                                                 Editor
Expires: August 4, 2002


            An RTP Payload Format for EVRC and SMV Vocoders


STATUS OF THIS MEMO

   This document is an Internet-Draft and is in full conformance with
   all provisions of Section 10 of RFC 2026.

   Internet-Drafts are working documents of the Internet Engineering
   Task Force (IETF), its areas, and its working groups. Note that other
   groups may also distribute working documents as Internet-Drafts.

   Internet-Drafts are draft documents valid for a maximum of six months
   and may be updated, replaced, or obsoleted by other documents at any
   time. It is inappropriate to use Internet- Drafts as reference
   material or to cite them other than as work in progress.

   The list of current Internet-Drafts can be accessed at
   http://www.ietf.org/ietf/1id-abstracts.txt

   The list of Internet-Draft Shadow Directories can be accessed at
   http://www.ietf.org/shadow.html.


ABSTRACT

   This document describes the RTP payload format for Enhanced Variable
   Rate Codec (EVRC) Speech and Selectable Mode Vocoder (SMV) Speech.
   Two sub-formats are specified for different application scenarios. A
   bundled/interleaved format is included to reduce the effect of packet
   loss on speech quality and amortize the overhead of the RTP header
   over more than one speech frame. A non-bundled format is also
   supported for conversational applications.


Table of Contents

   1. Introduction ................................................... 2
   2. Background ..................................................... 2
   3. The Codecs Supported ........................................... 3
   3.1. EVRC ......................................................... 3
   3.2. SMV .......................................................... 3
   3.3. Other Frame-Based Vocoders ................................... 4
   4. RTP/Vocoder Packet Format ...................................... 4



Adam H. Li                                                      [Page 1]


INTERNET-DRAFT   An RTP Payload Format for EVRC and SMV     Feb. 4, 2002


   4.1. Type 1 Interleaved/Bundled Packet Format ..................... 4
   4.2. Type 2 Header-Free Packet Format ............................. 6
   4.3. Detecting the Format of Packets .............................. 6
   5. Packet Table of Contents Entries and Codec Data Frame Format ... 7
   5.1. Packet Table of Contents entries ............................. 7
   5.2. Codec Data Frames ............................................ 8
   6. Interleaving Codec Data Frames in Type 1 Packets ............... 9
   6.1. Finding Interleave Group Boundaries ......................... 10
   6.2. Reconstructing Interleaved Speech ........................... 11
   6.3. Receiving Invalid Interleaving Values ....................... 12
   6.4. Additional Receiver Responsibilities ........................ 12
   7. Bundling Codec Data Frames in Type 1 Packets .................. 12
   8. Handling Missing Codec Data Frames ............................ 12
   9. Implementation Issues ......................................... 13
   9.1. Interleaving Length ......................................... 13
   9.2. Mode Request ................................................ 13
   10. IANA Considerations .......................................... 14
   10.1 Storage Mode ................................................ 14
   10.2 EVRC MIME Registration ...................................... 15
   10.3 SMV MIME Registration ....................................... 16
   11. Mapping to SDP Parameters .................................... 17
   12. Security Considerations ...................................... 17
   13. Adding Support of Other Frame-Based Vocoders ................. 18
   14. Acknowledgements ............................................. 18
   15. References ................................................... 18
   16. Authors' Address ............................................. 19


1. Introduction

   This document describes how speech compressed with EVRC [1] or SMV
   [2] may be formatted for use as an RTP payload type.  The format is
   also extensible to other codecs that generate a similar set of frame
   types. Two methods are provided to packetize the codec data frames
   into RTP packets: an interleaved/bundled format and a zero-header
   format. The sender may choose the best format for each application
   scenario, based on network conditions, bandwidth availability, delay
   requirements, and packet-loss tolerance.

   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 RFC 2119 [3].


2. Background

   The 3rd Generation Partnership Project 2 (3GPP2) has published two
   standards which define speech compression algorithms for CDMA
   applications: EVRC [1] and SMV [2]. EVRC is currently deployed in
   millions of first and second generation CDMA handsets. SMV is the
   preferred speech codec standard for CDMA2000, and will be deployed in
   third generation handsets in addition to EVRC. Improvements and new


Adam H. Li                                                      [Page 2]


INTERNET-DRAFT   An RTP Payload Format for EVRC and SMV     Feb. 4, 2002


   codecs will keep emerging as technology improves, and future handsets
   will likely support multiple codecs.

   The formats of the EVRC and SMV codec frames are very similar. Many
   other vocoders also share common characteristics, and have many
   similar application scenarios. This parallelism enables an RTP
   payload format to be designed for EVRC and SMV that may also support
   other, similar vocoders with minimal additional specification work.
   This can simplify the protocol for transporting vocoder data frames
   through RTP and reduce the complexity of implementations.


3. The Codecs Supported

3.1. EVRC

   The Enhanced Variable Rate Codec (EVRC) [1] compresses each 20
   milliseconds of 8000 Hz, 16-bit sampled speech input into output
   frames in one of the three different sizes: Rate 1 (171 bits), Rate
   1/2 (80 bits), or Rate 1/8 (16 bits). In addition, there are two zero
   bit codec frame types: null frames and erasure frames. Null frames
   are produced as a result of the vocoder running at rate 0. Null
   frames are zero bits long and are normally not transmitted. Erasure
   frames are the frames substituted by the receiver to the codec for
   the lost or damaged frames. Erasure frames are also zero bits long
   and are normally not transmitted.

   The codec chooses the output frame rate based on analysis of the
   input speech and the current operating mode (either normal or one of
   several reduced rate modes). For typical speech patterns, this
   results in an average output of 4.2 kilobits/second for normal mode
   and a lower average output for reduced rate modes.

3.2. SMV

   The Selectable Mode Vocoder (SMV) [2] compresses each 20 milliseconds
   of 8000 Hz, 16-bit sampled speech input into output frames of one of
   the four different sizes: Rate 1 (171 bits), Rate 1/2 (80 bits), Rate
   1/4 (40 bits), or Rate 1/8 (16 bits). In addition, there are two zero
   bit codec frame types: null frames and erasure frames. Null frames
   are produced as a result of the vocoder running at rate 0. Null
   frames are zero bits long and are normally not transmitted. Erasure
   frames are the frames substituted by the receiver to the codec for
   the lost or damaged frames. Erasure frames are also zero bits long
   and are normally not transmitted.

   The SMV codec can operate in four modes. Each mode may produce frames
   of any of the rates (full rate to 1/8 rate) for varying percentages
   of time, based on the characteristics of the speech samples and the
   selected mode. The SMV mode can change on a frame-by-frame basis. The
   SMV codec does not need additional information other than the codec
   data frames to correctly decode the data of various modes; therefore,


Adam H. Li                                                      [Page 3]


INTERNET-DRAFT   An RTP Payload Format for EVRC and SMV     Feb. 4, 2002


   the mode of the encoder does not need to be transmitted with the
   encoded frames.

   The percentage of different frame rates and the average data rate
   (ADR) for the four SMV modes are shown in the table below.

                     Mode 0       Mode 1       Mode 2        Mode 3
       -------------------------------------------------------------
       Rate 1        68.90%       38.14%       15.43%        07.49%
       Rate 1/2      06.03%       15.82%       38.34%        46.28%
       Rate 1/4      00.00%       17.37%       16.38%        16.38%
       Rate 1/8      25.07%       28.67%       29.85%        29.85%
       -------------------------------------------------------------
       ADR          7205 bps     5182 bps     4073 bps      3692 bps

   The SMV codec chooses the output frame rate based on an analysis of
   the input speech and the current operating mode. For typical speech
   patterns, this results in an average output of 4.2k bits/second for
   Mode 0 and lower for other reduced rate modes.

   SMV is more bandwidth efficient than EVRC. EVRC is equivalent in
   performance to SMV mode 1.

3.3. Other Frame-Based Vocoders

   Other frame-based vocoders can be carried in the packet format
   defined in this document, as long as they possess the following
   properties:

    o The codec is frame-based;
    o blank and erasure frames are supported;
    o the total number of rates is less than 17;
    o the maximum full rate frame can be transported in a single RTP
      packet using this specific format.

   Vocoders with the characteristics listed above can be transported
   using the packet format specified in this document with some
   additional specification work; the pieces that must be defined are
   listed in Section 13.


4. RTP/Vocoder Packet Format

   The RTP payload data MUST be transmitted in packets of one of the
   following two types.

4.1. Type 1 Interleaved/Bundled Packet Format

   This format is used to send one or more vocoder frames per packet.
   Interleaving or bundling MAY be used. The RTP packet for this format
   is as follows:



Adam H. Li                                                      [Page 4]


INTERNET-DRAFT   An RTP Payload Format for EVRC and SMV     Feb. 4, 2002


    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
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
   |                      RTP Header [4]                           |
   +=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+
   |R|R| LLL | NNN | FFF |  Count  |  TOC  |  ...  |  TOC  |padding|
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
   |        one or more codec data frames, one per TOC entry       |
   |                             ....                              |
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+

   The RTP header has the expected values as described in the RTP
   specification [4]. The RTP timestamp is in 1/8000 of a second units
   for EVRC and SMV. For any other vocoders that use this packet format,
   the timestamp unit needs to be defined explicitly. The M bit should
   be set as specified in the applicable RTP profile, for example, RFC
   1890 [5]. Note that RFC 1890 [5] specifies that if the sender does
   not suppress silence, the M bit will always be zero. When multiple
   codec data frames are present in a single RTP packet, the timestamp
   is, as always, that of the oldest data represented in the RTP packet.
   The assignment of an RTP payload type for this new packet format is
   outside the scope of this document, and will not be specified here.
   It is expected that the RTP profile for a particular class of
   applications will assign a payload type for this encoding, or if that
   is not done, then a payload type in the dynamic range shall be chosen
   by the sender.

   The first octet of a Type 1 Interleaved/Bundled format packet is the
   Interleave Octet. The second octet contains the Mode Request and
   Frame Count fields. The Table of Contents (ToC) field then follows.
   The fields are specified as follows:

   Reserved (RR): 2 bits
      Reserved bits. MUST be set to zero by sender, SHOULD be ignored
      by receiver.

   Interleave Length (LLL): 3 bits
      Indicates the length of interleave; a value of 0 indicates
      bundling, a special case of interleaving. See Section 6 and
      Section 7 for more detailed discussion.

   Interleave Index (NNN): 3 bits
      Indicates the index within an interleave group. MUST have a value
      less than or equal to the value of LLL. Values of NNN greater
      than the value of LLL are invalid. Packet with invalid NNN values
      SHOULD be ignored by the receiver.

   Mode Request (FFF): 3 bits
      The Mode Request field is used to signal Mode Request
      information. See Section 9.2 for details.




Adam H. Li                                                      [Page 5]


INTERNET-DRAFT   An RTP Payload Format for EVRC and SMV     Feb. 4, 2002


   Frame Count (Count): 5 bits
      Indicates the number of ToC fields (and therefore vocoder frames)
      present. A value of zero indicates that the packet contains one
      ToC field (and vocoder frame). A value of 31 indicates 32 ToC
      fields (and vocoder frames) are in the packet. The number of ToC
      fields (and vocoder frames) present is the value of the frame
      count field plus one.

   Padding (padding): 0 or 4 bits
      This padding ensures that codec data frames start on an octet
      boundary. When the frame count is odd, the sender MUST add 4 bits
      of padding following the last TOC. When the frame count is even,
      the sender MUST NOT add padding bits. If padding is present, the
      padding bits MUST be set to zero by sender, and SHOULD be ignored
      by receiver.

   The Table of Contents field (ToC) provides information on the codec
   data frame(s) in the packet. There is one ToC entry for each codec
   data frame. The detailed formats of the ToC field and codec data
   frames are specified in Section 5.

   Multiple data frames may be included within a Type 1
   Interleaved/Bundled packet using interleaving or bundling as
   described in Section 6 and Section 7.

4.2. Type 2 Header-Free Packet Format

   The Type 2 Header-Free Packet Format is designed for maximum
   bandwidth efficiency and low latency. Only one codec data frame can
   be sent in each Type 2 Header-Free format packet. None of the payload
   header fields (LLL, NNN, FFF, Count) nor ToC entries are present. The
   codec rate for the data frame can be determined from the length of
   the codec data frame, since there is only one codec data frame in
   each Type 2 Header-Free packet.

   Use of the RTP header fields for Type 2 Header-Free RTP/Vocoder
   Packet Format is the same as described in Section 4.1 for Type 1
   Interleaved/Bundled RTP/Vocoder Packet Format. The detailed format of
   the codec data frame is specified in Section 5.

    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
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
   |                      RTP Header [4]                           |
   +=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+
   |                                                               |
   +          ONLY one codec data frame            +-+-+-+-+-+-+-+-+
   |                                               |
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+





Adam H. Li                                                      [Page 6]


INTERNET-DRAFT   An RTP Payload Format for EVRC and SMV     Feb. 4, 2002


4.3. Detecting the Format of Packets

   All receivers MUST be able to process both types of packets. The
   sender MAY choose to use one or both types of packets.

   A receiver MUST have prior knowledge of the packet type to correctly
   decode the RTP packets. The packet types used in an RTP session MUST
   be specified by the sender, and signaled through out-of-band means,
   for example by SDP during the setup of a session.

   When packets of both formats are used within the same session,
   different RTP payload type values MUST be used for each format to
   distinguish the packet formats. The association of payload type
   number with the packet format is done out-of-band, for example by SDP
   during the setup of a session.


5. Packet Table of Contents Entries and Codec Data Frame Format

5.1. Packet Table of Contents entries

   Each codec data frame in a Type 1 Interleaved/Bundled packet has a
   corresponding Table of Contents (ToC) entry. The ToC entry indicates
   the rate of the codec frame. (Type 2 Header-Free packets MUST NOT
   have a ToC field, and there is always only one codec data frame in
   each Type 2 Header-Free packet.)

   Each ToC entry is occupies four bits. The format of the bits is
   indicated below:

       0 1 2 3
      +-+-+-+-+
      |fr type|
      +-+-+-+-+

   Frame Type: 4 bits
      The frame type indicates the type of the corresponding codec data
      frame in the RTP packet.

      For EVRC and SMV codecs, the frame type values and size of the
      associated codec data frame are described in the table below:

      Value   Rate      Total codec data frame size (in octets)
      ---------------------------------------------------------
        0     Blank      0    (0 bit)
        1     1/8        2    (16 bits)
        2     1/4        5    (40 bits; not valid for EVRC)
        3     1/2       10    (80 bits)
        4     1         22    (171 bits; 5 padded at end with zeros)
        5     Erasure    0    (SHOULD NOT be transmitted by sender)

      All values not listed in the above table MUST be considered
      reserved. A ToC entry with a reserved Frame Type value SHOULD be

Adam H. Li                                                      [Page 7]


INTERNET-DRAFT   An RTP Payload Format for EVRC and SMV     Feb. 4, 2002


      considered invalid and substituted with an erasure frame. Note
      that the EVRC codec does not have 1/4 rate frames, thus frame
      type value 2 MUST be considered a reserved value when the EVRC
      codec is in use.

      Other vocoders that use this packet format need to specify their
      own table of frame types and corresponding codec data frames.

5.2. Codec Data Frames

   The output of the vocoder MUST be converted into codec data frames
   for inclusion in the RTP payload. The conversions for EVRC and SMV
   codecs are specified below. (Note: Because the EVRC codec does not
   have Rate 1/4 frames, the specifications of 1/4 frames does not apply
   to EVRC codec data frames). Other vocoders that use this packet
   format need to specify how to convert vocoder output data into
   frames.

   The codec output data bits as numbered in EVRC and SMV are packed
   into octets. The lowest numbered bit (bit 1 for Rate 1, Rate 1/2,
   Rate 1/4 and Rate 1/8) is placed in the most significant bit
   (internet bit 0) of octet 1 of the codec data frame, the second
   lowest bit is placed in the second most significant bit of the first
   octet, the third lowest in the third most significant bit of the
   first octet, and so on. This continues until all of the bits have
   been placed in the codec data frame.

   The remaining unused bits of the last octet of the codec data frame
   MUST be set to zero. Note that in EVRC and SMV this is only
   applicable to Rate 1 frames (171 bits) as the Rate 1/2 (80 bits),
   Rate 1/4 (40 bits, SMV only) and Rate 1/8 frames (16 bits) fit
   exactly into a whole number of octets.

   Following is a detailed listing showing a Rate 1 EVRC/SMV codec
   output frame converted into a codec data frame:

   The codec data frame for a EVRC/SMV codec Rate 1 frame is 22 octets
   long. Bits 1 through 171 from the EVRC/SMV codec Rate 1 frame are
   placed as indicated, with bits marked with "Z" set to zero. EVRC/SMV
   codec Rate 1/8, Rate 1/4 and Rate 1/2 frames are converted similarly,
   but do not require zero padding because they align on octet
   boundaries.

                    Rate 1 codec data frame (octets 0 - 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
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
   |0|0|0|0|0|0|0|0|0|0|0|0|0|0|0|0|0|0|0|0|0|0|0|0|0|0|0|0|0|0|0|0|
   |0|0|0|0|0|0|0|0|0|1|1|1|1|1|1|1|1|1|1|2|2|2|2|2|2|2|2|2|2|3|3|3|
   |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|2|
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+


Adam H. Li                                                      [Page 8]


INTERNET-DRAFT   An RTP Payload Format for EVRC and SMV     Feb. 4, 2002


                    Rate 1 codec data frame (octets 19 - 21)

    1           1                   1                   1
    4           5                   6                   7
    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 2 3 4 5
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
   |1|1|1|1|1|1|1|1|1|1|1|1|1|1|1|1|1|1|1|1|1|1|1|1|1|1|1| | | | | |
   |4|4|4|4|4|5|5|5|5|5|5|5|5|5|5|6|6|6|6|6|6|6|6|6|6|7|7|Z|Z|Z|Z|Z|
   |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| | | | | |
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+


6. Interleaving Codec Data Frames in Type 1 Packets

   As indicated in Section 4.1, more than one codec data frame MAY be
   included in a single Type 1 Interleaved/Bundled packet by a sender.
   This is accomplished by interleaving or bundling.

   Bundling is used to spread the transmission overhead of the RTP and
   payload header over multiple vocoder frames. Interleaving
   additionally reduces the listener's perception of data loss by
   spreading such loss over non-consecutive vocoder frames. EVRC, SMV,
   and similar vocoders are able to compensate for an occasional lost
   frame, but speech quality degrades exponentially with consecutive
   frame loss.

   Bundling is signaled by setting the LLL field to zero and the Count
   field to greater than zero. Interleaving is indicated by setting the
   LLL field to a value greater than zero.

   The discussions on general interleaving apply to the bundling (which
   can be viewed as a reduced case of interleaving) with reduced
   complexity. The bundling case is discussed in detail in Section 7.

   Senders MAY support interleaving and/or bundling. All receivers MUST
   support interleaving and bundling.

   Given a time-ordered sequence of output frames from the EVRC codec
   numbered 0..n, a bundling value B (in the Count field), and an
   interleave length L where n = B * (L+1) - 1, the output frames are
   placed into RTP packets as follows (the values of the fields LLL and
   NNN are indicated for each RTP packet):

   First RTP Packet in Interleave group:
      LLL=L, NNN=0
      Frame 0, Frame L+1, Frame 2(L+1), Frame 3(L+1), ... for a total of
      B frames

   Second RTP Packet in Interleave group:
      LLL=L, NNN=1
      Frame 1, Frame 1+L+1, Frame 1+2(L+1), Frame 1+3(L+1), ... for a
      total of B frames


Adam H. Li                                                      [Page 9]


INTERNET-DRAFT   An RTP Payload Format for EVRC and SMV     Feb. 4, 2002


   This continues to the last RTP packet in the interleave group:

   L+1 RTP Packet in Interleave group:
      LLL=L, NNN=L
      Frame L, Frame L+L+1, Frame L+2(L+1), Frame L+3(L+1), ... for a
      total of B frames

   Within each interleave group, the RTP packets making up the
   interleave group MUST be transmitted in value-increasing order of the
   NNN field. While this does not guarantee reduced end-to-end delay on
   the receiving end, when packets are delivered in order by the
   underlying transport, delay will be reduced to the minimum possible.

   Receivers MAY signal the maximum number of codec data frames (i.e.,
   the maximum acceptable bundling value B) they can handle in a single
   RTP packet using the OPTIONAL maxptime RTP mode parameter identified
   in Section 10.

   Receivers MAY signal the maximum interleave length (i.e., the maximum
   acceptable LLL value in the Interleaving Octet) they will accept
   using the OPTIONAL maxinterleave RTP mode parameter identified in
   Section 10.

   Additionally, senders have the following restrictions:

   o  MUST NOT bundle more codec data frames in a single RTP packet than
      indicated by maxptime (see Section 10) if it is signaled.

   o  SHOULD NOT bundle more codec data frames in a single RTP packet
      than will fit in the MTU of the underlying network.

   o  Once beginning a session with a given maximum interleaving value
      set by maxinterleave in Section 10, MUST NOT increase the
      interleaving value (LLL) to exceed the maximum interleaving value
      that is signaled.

   o  MAY change the interleaving value only between interleave groups.

   o  Silence suppression MAY only be used between interleave groups. A
      ToC with Frame Type 0 (Blank Frame, Section 5.1) MUST be used
      within interleaving groups if the codec outputs a blank frame.
      The M bits in the RTP header MUST NOT be set, as the stream is
      continuous in time. Because there is only one time stamp for each
      RTP packet, silence suppression used within an interleave group
      will cause ambiguities when reconstructing the speech at the
      receiver side, and thus is prohibited.

6.1. Finding Interleave Group Boundaries

   Given an RTP packet with sequence number S, interleave length (field
   LLL) L, interleave index value (field NNN) N, and bundling value B,
   the interleave group consists of this RTP packet and other RTP
   packets with sequence numbers from S-N to S-N+L inclusive. (The

Adam H. Li                                                     [Page 10]


INTERNET-DRAFT   An RTP Payload Format for EVRC and SMV     Feb. 4, 2002


   sequence numbers used here are for illustrative purposes. When
   wrapping around happens, the sequence numbers need to be adjusted
   accordingly). In other words, the interleave group always consists of
   L+1 RTP packets with sequential sequence numbers. The bundling value
   for all RTP packets in an interleave group MUST be the same.

   The receiver determines the expected bundling value for all RTP
   packets in an interleave group by the number of codec data frames
   bundled in the first RTP packet of the interleave group received.
   Note that this may not be the first RTP packet of the interleave
   group if packets are delivered out of order by the underlying
   transport.

   On receipt of an RTP packet in an interleave group with other than
   the expected bundling value, the receiver MAY discard codec data
   frames off the end of the RTP packet or add erasure codec data frames
   to the end of the packet in order to manufacture a substitute packet
   with the expected bundling value.  The receiver MAY instead choose to
   discard the whole interleave group.

6.2. Reconstructing Interleaved Speech

   Given an RTP sequence number ordered set of RTP packets in an
   interleave group numbered 0..L, where L is the interleave length and
   B is the bundling value, and codec data frames within each RTP packet
   that are numbered in order from first to last with the numbers 1..B,
   the original, time-ordered sequence of output frames from the EVRC
   codec may be reconstructed as follows:

   First L+1 frames:
      Frame 0 from packet 0 of interleave group
      Frame 0 from packet 1 of interleave group
      And so on up to...
      Frame 0 from packet L of interleave group

   Second L+1 frames:
      Frame 1 from packet 0 of interleave group
      Frame 1 from packet 1 of interleave group
      And so on up to...
      Frame 1 from packet L of interleave group

   And so on up to...

   Bth L+1 frames:
      Frame B from packet 0 of interleave group
      Frame B from packet 1 of interleave group
      And so on up to...
      Frame B from packet L of interleave group






Adam H. Li                                                     [Page 11]


INTERNET-DRAFT   An RTP Payload Format for EVRC and SMV     Feb. 4, 2002


6.3. Receiving Invalid Interleaving Values

   On receipt of an RTP packet with an invalid value of the LLL or NNN
   fields, the RTP packet SHOULD be treated as lost by the receiver for
   the purpose of generating erasure frames as described in Section 8.

6.4. Additional Receiver Responsibilities

   Assume that the receiver has begun playing frames from an interleave
   group. The time has come to play frame x from packet n of the
   interleave group. Further assume that packet n of the interleave
   group has not been received. As described in section 8, an erasure
   frame will be sent to the receiving vocoder.

   Now, assume that packet n of the interleave group arrives before
   frame x+1 of that packet is needed. Receivers SHOULD use frame x+1 of
   the newly received packet n rather than substituting an erasure
   frame. In other words, just because packet n was not available the
   first time it was needed to reconstruct the interleaved speech, the
   receiver SHOULD NOT assume it is not available when it is
   subsequently needed for interleaved speech reconstruction.


7. Bundling Codec Data Frames in Type 1 Packets

   As discussed in Section 6, the bundling of codec data frames is a
   special reduced case of interleaving with LLL value in the Interleave
   Octet set to 0.

   Bundling codec data frames indicates multiple data frames are
   included consecutively in a packet, because the interleaving length
   (LLL) is 0. The interleaving group is thus reduced to a single RTP
   packet, and the reconstruction of the code data frames from RTP
   packets becomes a much simpler process.

   Furthermore, the additional restrictions on senders are reduced to:

   o  MUST NOT bundle more codec data frames in a single RTP packet than
      indicated by maxptime (see Section 10) if it is signaled.

   o  SHOULD NOT bundle more codec data frames in a single RTP packet
      than will fit in the MTU of the underlying network.


8. Handling Missing Codec Data Frames

   The vocoders covered by this payload format support erasure frame as
   an indication when frames are not available. While an erasure frame
   MUST NOT be transmitted by an RTP sender, it MAY be used internally
   by a receiver to advance the state of the voice decoder by exactly
   one frame time for each missing frame. Using the information from
   packet sequence number, time stamp, and the M bit, the receiver can
   detect missing codec data frames from RTP packet loss and/or silence

Adam H. Li                                                     [Page 12]


INTERNET-DRAFT   An RTP Payload Format for EVRC and SMV     Feb. 4, 2002


   suppression, and generate corresponding erasure frames. Erasure
   frames SHOULD also be used in storage mode to record missing frames.


9. Implementation Issues

9.1. Interleaving Length

   The vocoder interpolates the missing speech content when given an
   erasure frame. However, the best quality is perceived by the listener
   when erasure frames are not consecutive. This makes interleaving
   desirable as it increases speech quality when packet loss occurs.

   On the other hand, interleaving can greatly increase the end-to-end
   delay. Where an interactive session is desired, either Type 1
   Interleaved/Bundled with interleaving length (field LLL) 0 or Type 2
   Header-Free RTP payload types are RECOMMENDED.

   When end-to-end delay is not a concern, an interleaving length (field
   LLL) of 4 or 5 is RECOMMENDED.

   The parameters maxptime and maxinterleave are exchanged at the
   initial setup of the session so that the receiver can allocate a
   known amount of buffer space that will be sufficient for all future
   reception in that session. During the session, the sender may
   decrease the bundling value or interleaving length (so that less
   buffer space is required at the receiver), but never require more
   buffer space. This prevents the situation where a receiver needs to
   allocate more buffer space in the middle of a session but is unable
   to do so.

9.2. Mode Request

   The Mode Request signal requests a particular encoding mode for the
   speech encoding in the reverse direction. All implementations are
   RECOMMENDED to honor the Mode Request signal. The Mode Request signal
   SHOULD only be used in one-to-one sessions. In multiparty sessions,
   any received Mode Request signals SHOULD be ignored.

   In addition, the Mode Request signal MAY also be sent through non-RTP
   means, which is out of the scope of this specification.

   The three-bit Mode Request field is used to signal the receiver to
   set a particular encoding mode to its audio encoder. If the Mode
   Request field is set to a non-zero value in RTP packets from node A
   to node B, it is a request for node B to change to the requested
   encoding mode for its audio encoder and therefore the bit rate of the
   RTP stream from node B to node A. Once a node sets this field to a
   non-zero value it SHOULD continue to set the field to the same value
   in subsequent packets until the requested mode has changed. This
   design helps to eliminate the scenario of getting the codec stuck in
   an unintended state if one of the packets that carries the Mode


Adam H. Li                                                     [Page 13]


INTERNET-DRAFT   An RTP Payload Format for EVRC and SMV     Feb. 4, 2002


   Request is lost. An otherwise silent node MAY send an RTP packet
   containing a blank frame in order to send a Mode Request.

   Each codec type using this format SHOULD define its own
   interpretation of the Mode Request field. Codecs SHOULD follow the
   convention that higher values of the three-bit field correspond to an
   equal or lower average output bit rate.

   For the EVRC codec, the Mode Request field MUST be interpreted
   according to Tables 2.2.1.2-1 and 2.2.1.2-2 of the EVRC codec
   specifications [1].  Values above '100' (4) are currently reserved.
   If an unknown value above '100' (4) is received, it MUST be handled
   as if '100' (4) were received.

   For SMV codec, the Mode Request field MUST be interpreted according
   to Table 2.2-2 of the SMV codec specifications [2]. Values above
   '101' (5) are currently reserved. If an unknown value above '101' (5)
   is received, it MUST be handled as if '101' (5) were received.


10. IANA Considerations

   Two new MIME sub-types as described in this section are to be
   registered.

   The MIME-names for the EVRC and SMV codec are allocated from the IETF
   tree since all the vocoders covered are expected to be widely used
   for Voice-over-IP applications.

   The RTP mode has been described in the previous sections.

10.1. Storage Mode

   The storage mode is used for storing speech frames, e.g., as a file
   or e-mail attachment.

   The file begins with a magic number to identify the vocoder that is
   used. The magic number for EVRC corresponds to the ASCII character
   string "#!EVRC\n", i.e., "0x23 0x21 0x45 0x56 0x52 0x43 0x0A" in
   network byte order. The magic number for SMV corresponds to the ASCII
   character string "#!SMV\n", i.e., "0x23 0x21 0x53 0x4d 0x56 0x0a" in
   network byte order.

   The codec data frames are stored in consecutive order, with a single
   TOC entry field, expanded to one octet, prefixing each codec data
   frame. The ToC field is expanded to one octet by setting the left-
   most four bits of the octet to zero. For example, a ToC value of 4 (a
   full-rate frame) is stored as 0x04.

   Speech frames lost in transmission and non-received frames MUST be
   stored as erasure frames (frame type 5, see definition in Section
   5.1) to maintain synchronization with the original media.


Adam H. Li                                                     [Page 14]


INTERNET-DRAFT   An RTP Payload Format for EVRC and SMV     Feb. 4, 2002


10.2. EVRC MIME Registration

   Media Type Name:     audio

   Media Subtype Name:  EVRC

   Required Parameter for RTP mode:

      ptype:    Indicates the Type of the RTP/Vocoder packets. The
         valid values are 1 (Type 1 Interleaved/Bundled) or 2 (Type 2
         Header-Free).

   Optional parameters for RTP mode:

      ptime:    Defined as usual for RTP audio [6].

      maxptime: The maximum amount of media which can be encapsulated
         in each packet, expressed as time in milliseconds. The time
         SHALL be calculated as the sum of the time the media present
         in the packet represents. The time SHOULD be a multiple of the
         duration of a single codec data frame (20 msec). If not
         signaled, the default maxptime value SHALL be 200
         milliseconds.

      maxinterleave: Maximum number for interleaving length (field LLL
         in the Interleaving Octet). The interleaving lengths used in
         the entire session MUST NOT exceed this maximum value. If not
         signaled, the maxinterleave length SHALL be 5.

   Optional parameters for storage mode: none

   Encoding considerations for RTP mode: see Section 6 and Section 7 of
      RFC xxxx.

   Encoding considerations for storage mode: see Section 10.1 of RFC
      xxxx.

   Security considerations: see Section 12 "Security Considerations" of
      RFC xxxx.

   Public specification: RFC xxxx.

   Additional information for storage mode:
      Magic number: #!EVRC\n
      File extensions: evc, EVC
      Macintosh file type code: none
      Object identifier or OID: none

   Intended usage: COMMON. It is expected that many VoIP applications
      (as well as mobile applications) will use this type.

   Person & email address to contact for further information:
      Adam Li

Adam H. Li                                                     [Page 15]


INTERNET-DRAFT   An RTP Payload Format for EVRC and SMV     Feb. 4, 2002


      adamli@icsl.ucla.edu

   Author/Change controller:
      Adam Li
      adamli@icsl.ucla.edu
      IETF Audio/Video Transport Working Group

10.3. SMV MIME Registration

   Media Type Name:     audio

   Media Subtype Name:  SMV

   Required Parameter for RTP mode:

      ptype:    Indicates the Type of the RTP/Vocoder packets. The
         valid values are 1 (Type 1 Interleaved/Bundled) or 2 (Type 2
         Header-Free).

   Optional parameters for RTP mode:

      ptime:    Defined as usual for RTP audio [6].

      maxptime: The maximum amount of media which can be encapsulated
         in each packet, expressed as time in milliseconds. The time
         SHALL be calculated as the sum of the time the media present
         in the packet represents. The time SHOULD be a multiple of the
         duration of a single codec data frame (20 msec). If not
         signaled, the default maxptime value SHALL be 200
         milliseconds.

      maxinterleave: Maximum number for interleaving length (field LLL
         in the Interleaving Octet). The interleaving lengths used in
         the entire session MUST NOT exceed this maximum value. If not
         signaled, the maxinterleave length SHALL be 5.

   Optional parameters for storage mode: none

   Encoding considerations for RTP mode: see Section 6 and Section 7 of
      RFC xxxx.

   Encoding considerations for storage mode: see Section 10.1 of RFC
      xxxx.

   Security considerations: see Section 12 "Security Considerations" of
      RFC xxxx.

   Public specification: RFC xxxx.

   Additional information for storage mode:
      Magic number: #!SMV\n
      File extensions: smv, SMV
      Macintosh file type code: none

Adam H. Li                                                     [Page 16]


INTERNET-DRAFT   An RTP Payload Format for EVRC and SMV     Feb. 4, 2002


      Object identifier or OID: none

   Intended usage: COMMON. It is expected that many VoIP applications
      (as well as mobile applications) will use this type.

   Person & email address to contact for further information:
      Adam Li
      adamli@icsl.ucla.edu

   Author/Change controller:
      Adam Li
      adamli@icsl.ucla.edu
      IETF Audio/Video Transport Working Group


11. Mapping to SDP Parameters

   Please note that this section applies to the RTP mode only.

   Parameters are mapped to SDP [6] as usual.
   Example usage in SDP:
     m = audio 49120 RTP/AVP 97
     a = rtpmap:97 EVRC
     a = fmtp:97 ptype=1; maxinterleave=2
     a = maxptime:80


12. Security Considerations

   RTP packets using the payload format defined in this specification
   are subject to the security considerations discussed in the RTP
   specification [4], and any appropriate profile (for example [5]).
   This implies that confidentiality of the media streams is achieved by
   encryption. Because the data compression used with this payload
   format is applied end-to-end, encryption may be performed after
   compression so there is no conflict between the two operations.

   A potential denial-of-service threat exists for data encoding using
   compression techniques that have non-uniform receiver-end
   computational load. The attacker can inject pathological datagrams
   into the stream which are complex to decode and cause the receiver to
   become overloaded. However, the encodings covered in this document do
   not exhibit any significant non-uniformity.

   As with any IP-based protocol, in some circumstances, a receiver may
   be overloaded simply by the receipt of too many packets, either
   desired or undesired. Network-layer authentication may be used to
   discard packets from undesired sources, but the processing cost of
   the authentication itself may be too high. In a multicast
   environment, pruning of specific sources may be implemented in
   future versions of IGMP [7] and in multicast routing protocols to
   allow a receiver to select which sources are allowed to reach it.


Adam H. Li                                                     [Page 17]


INTERNET-DRAFT   An RTP Payload Format for EVRC and SMV     Feb. 4, 2002


   Interleaving MAY affect encryption. Depending on the used encryption
   scheme there MAY be restrictions on for example the time when keys
   can be changed.


13. Adding Support of Other Frame-Based Vocoders

   As described above, the RTP packet format defined in this document is
   very flexible and designed to be usable by other frame-based
   vocoders.

   Additional vocoders using this format MUST have properties as
   described in Section 3.3.

   The following need to be done in order for any eligible vocoders to
   use the RTP payload format defined in this document:

    o Define the unit used for RTP time stamp;
    o Define the meaning of the Mode Request bits;
    o Define corresponding codec data frame type values for ToC;
    o Define the conversion procedure for vocoders output data frame;
    o Define a magic number for storage mode, and complete the
      corresponding MIME registration.


14. Acknowledgements

   The following authors have made significant contributions to this
   document: Adam H. Li, John D. Villasenor, Dong-Seek Park, Jeong-Hoon
   Park, Keith Miller, S. Craig Greer, David Leon, Nikolai Leung,
   Marcello Lioy, Kyle J. McKay, Magdalena L. Espelien, Randall Gellens,
   Tom Hiller, Peter J. McCann, Stinson S. Mathai, Michael D. Turner,
   Ajay Rajkumar, Dan Gal, Magnus Westerlund, Lars-Erik Jonsson, Greg
   Sherwood, and Thomas Zeng.


15. References

   [1]  3GPP2 C.S0014, "Enhanced Variable Rate Codec, Speech Service
        Option 3 for Wideband Spread Spectrum Digital Systems", January
        1997.

   [2]  3GPP2 C.S0030, "Selectable Mode Vocoder", August 2001.

   [3]  Bradner, S., "Key words for use in RFCs to Indicate Requirement
        Levels", BCP 14, RFC 2119, March 1997.

   [4]  Schulzrinne, H., Casner, S., Frederick, R. and V. Jacobson,
        "RTP:  A Transport Protocol for Real-Time Applications", RFC
        1889, January 1996.

   [5]  Schulzrinne, H., "RTP Profile for Audio and Video Conferences
        with Minimal Control", RFC 1890, January 1996.

Adam H. Li                                                     [Page 18]


INTERNET-DRAFT   An RTP Payload Format for EVRC and SMV     Feb. 4, 2002



   [6]  M. Handley and V. Jacobson, "SDP: Session Description Protocol",
        RFC 2327, April 1998.

   [7]  Deering, S., "Host Extensions for IP Multicasting", STD 5, RFC
        1112, August 1989.


16. Authors' Address

   The editor will serve as the point of contact for technical issues.

   Adam H. Li
   Image Communication Lab
   Electrical Engineering Department
   University of California
   Los Angeles, CA 90095
   USA
   Phone: +1 310 825 5178
   Email: adamli@icsl.ucla.edu


































Adam H. Li                                                     [Page 19]