RTP Payload for Comfort

This document describes an RTP [2] payload format for transporting comfort noise (CN). The CN payload type is primarily for use with audio codecs that do not support comfort noise as part of the codec itself such as ITU-T Recommendations G.711 [3], G.726 [4], G.727 [5], G.728 [6], and G.722 [7]. 1. Conventions used in this document 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 [8]. 2. Introduction This document describes an RTP payload format for transporting comfort noise. The payload format is based on Appendix II of ITU-T Recommendation G.711 [9] which defines a comfort noise payload format (or bit-stream) for ITU-T G.711 use in packet-based multimedia communication systems. The payload format is generic and may also be used with other audio codecs without built-in Zopf Standards Track July, 2002 1 RTP Payload for Comfort Noise January 2002 Discontinuous Transmission (DTX) capability such as ITU-T Recommendations G.726 [4], G.727 [5], G.728 [6], and G.722 [7]. The payload format provides a minimum interoperability specification for communication of comfort noise parameters. The comfort noise analysis and synthesis as well as the Voice Activity Detection (VAD) and DTX algorithms are unspecified and left implementation-specific. However, an example solution for G.711 has been tested and is described in the Appendix [9]. It uses the VAD and DTX of G.729 Annex B [10] and a comfort noise generation algorithm (CNG) which is provided in the Appendix for information. The comfort noise payload consists of a single octet description of the noise level and MAY contain spectral information in subsequent octets. An earlier version of the CN payload format consisting only of the noise level byte was defined in draft revisions of the RFC 1890. The extended payload format defined in this document should be backward compatible with implementations of the earlier version assuming that only the first byte is interpreted and any additional spectral information bytes are ignored. 3. CN Payload Definition The comfort noise payload consists of a description of the noise level and spectral information in the form of reflection coefficients. The use of spectral information is optional and the all-pole model order is left unspecified. The encoder can determine the appropriate model order based on such considerations as quality, complexity, expected environmental noise, and signal bandwidth. The model order is not explicitly transmitted since it can be derived from the length of the payload at the receiver. For complexity or other reasons, the decoder may reduce the model order by setting higher order reflection coefficients to zero. 3.1 Noise Level The magnitude of the noise level is packed into the least significant bits of the noise-level byte with the most significant bit unused and always set to 0 as shown below in Figure 1. The least significant bit of the noise level magnitude is packed into the least significant bit of the byte. The noise level is expressed in -dBov, with values from 0 to 127 representing 0 to -127 dBov. dBov is the level relative to the overload of the system. (Note: Representation relative to the overload point of a system is particularly useful for digital implementations, since one does not need to know the relative calibration of the analog circuitry.) For example, in the case of a u-law system, the reference would be a square wave with values +/8031, and this square wave represents 0dBov. This translates into 6.18dBm0. Zopf Standards Track July, 2002 2 RTP Payload for Comfort Noise January 2002 0 1 2 3 4 5 6 7 +-+-+-+-+-+-+-+-+ |0| level | +-+-+-+-+-+-+-+-+ Figure 1: Noise Level Packing 3.2 Spectral Information The spectral information is transmitted using reflection coefficients [9]. Each reflection coefficient can have values between -1 and 1 and is quantized uniformly using 8 bits. The quantized value is represented by the 8 bit index N, where N=0..,254, and index N=255 is reserved for future use. Each index N is packed into a separate byte with the MSB first. The quantized value of each reflection coefficient, k_i, can be obtained from its corresponding index using: k_i(N_i) = 258*(N_i-127) for N_i = 0...254; -1 < k_i < 1 ------------32768 3.3 Payload Packing The first byte of the payload MUST contain the noise level as shown in Figure 1. Quantized reflection coefficients are packed in subsequent bytes in ascending order as in Figure 2 where M is the model order. The total length of the payload is M+1 bytes. Note that a 0th order model (i.e. no spectral envelope information) reduces to transmitting only the energy level. Byte 1 2 3 ... M+1 +-----+-----+-----+-----+-----+ |level| N1 | N2 | ... | NM | +-----+-----+-----+-----+-----+ Figure 2: CN Payload Packing Format 4. Usage of RTP The RTP header for the comfort noise packet SHOULD be constructed as if the comfort noise were an independent codec. Thus, the RTP timestamp designates the beginning of the comfort noise period. When this payload format is used under the RTP profile specified in RFC 1890[11], a static payload type of 13 is assigned for RTP timestamp clock rate of 8,000 Hz; if other rates are needed, they MUST be defined through dynamic payload types. The RTP packet SHOULD NOT have the marker bit set. Each RTP packet containing comfort noise MUST contain exactly one CN payload per channel. This is required since the CN payload has a Zopf Standards Track July, 2002 3 RTP Payload for Comfort Noise January 2002 variable length. If multiple audio channels are used, each channel MUST use the same spectral model order ’M’. 5. Guidelines for Use An audio codec with DTX capabilities generally includes VAD, DTX, and CNG algorithms. The job of the VAD is to discriminate between active and inactive voice segments in the input signal. During inactive voice segments, the role of the CNG is to sufficiently describe the ambient noise while minimizing the transmission rate. A Silence Insertion Descriptor (SID) frame containing a description of the noise is packed into the CN payload and sent to the receiver to drive the CNG. The DTX algorithm determines when a SID frame is transmitted. During active voice segments, packets of the voice codec are transmitted and indicated in the RTP header by the static or dynamic payload type for that codec. At the beginning of an inactive voice segment (silence period), a SID frame packed into a CN packet is transmitted in the same RTP stream and indicated by the CN payload type. The SID frame update rate is left implementation specific. For example, the SID frame may be sent periodically or only when there is a significant change in the background noise characteristics. The CNG algorithm at the receiver uses the information in the SID to update its noise generation model and then produce an appropriate amount of comfort noise. The CN payload format provides a minimum interoperability specification for communication of comfort noise parameters. The comfort noise analysis and synthesis as well as the VAD and DTX algorithms are unspecified and left implementation-specific. However, an example solution for G.711 has been tested and is described in Appendix II of ITU-T Recommendation G.711 [9]. It uses the VAD and DTX of G.729 Annex B [10] and a comfort noise generation algorithm (CNG), which is provided in the Appendix for information. Additional guidelines for use such as the factors affecting system performance in the design of the VAD/DTX/CNG algorithms are described in the Appendix. 5.1 Usage of SDP When using the Session Description Protocol (SDP) [12] to specify RTP payload information, the use of comfort noise is indicated by the inclusion of a payload type for CN on the media description line. When using CN with the RTP/AVP profile [11] and a codec whose RTP timestamp clock rate is 8000 Hz, such as G.711 (PCMU, static payload type 0), the static payload type 13 for CN can be used: m=audio 49230 RTP/AVP 0 13 When using CN with a codec that has a different RTP timestamp clock rate, a dynamic payload type mapping (rtpmap attribute) is required. Zopf Standards Track July, 2002 4 RTP Payload for Comfort Noise January 2002 This example shows CN used with the G.722.1 codec (see RFC 3047 [13]): m=audio 49230 RTP/AVP 101 102 a=rtpmap:101 G7221/16000 a=fmtp:121 bitrate=24000 a=rtpmap:102 CN/16000 Omission of a payload type for CN on the media description line implies that comfort noise will not be used, but it does not imply that silence will not be suppressed. Discontinuous transmission (silence suppression) is indicated by an increment in the RTP timestamp without a corresponding increment in the RTP sequence number. 6. IANA Considerations This section defines a new RTP payload name and associated MIME type, CN (audio/CN). The payload format specified in this document is also assigned payload type 13 in the RTP Payload Types table of the RTP Parameters registry maintained by the Internet Assigned Numbers Authority (IANA). [Note to IANA and the RFC Editor, to be deleted before publication of this draft as an RFC: The RTP Payload Types table referenced in the previous paragraph is "closed" per the policy change established in draft-ietf-avt-profile-new-12, but that draft also says "Payload type 13 is reserved for a comfort noise payload format to be specified in a separate RFC." This document is intended to be that separate RFC.] 6.1 Registration of MIME media type audio/CN MIME media type name: audio MIME subtype name: CN Required parameters: None Optional parameters: rate: specifies the RTP timestamp clock rate, which is usually (but not always) equal to the sampling rate. This parameter should have the same value as the codec used in conjunction with comfort noise. The default value is 8000. Encoding considerations: This type is only defined for transfer via RTP [RFC 1889]. Security considerations: see Section 8 "Security Considerations". Interoperability considerations: none Published specification: This document and Appendix II of ITU-T Recommendation G.711 Zopf Standards Track July, 2002 5 RTP Payload for Comfort Noise January 2002 Applications which use this media type:Audio and video streaming and conferencing tools.Additional information: nonePerson & email address to contact for further information:Robert [email protected] usage: COMMONAuthor/Change controller:Author: Robert ZopfChange controller: IETF AVT Working Group7. Security ConsiderationsRTP packets using the payload format defined in this specificationare subject to the security considerations discussed in the RTPspecification [2]. This implies that confidentiality of the mediastreams is achieved by encryption. Because the payload format isarranged end-to-end, encryption MAY be performed after encapsulationso there is no conflict between the two operations.As this format transports background noise, there are no significantsecurity, confidentiality, or authentication concerns.8. References 1 Bradner, S., "The Internet Standards Process -Revision 3", BCP9, RFC 2026, October 1996.2 H. Schulzrinne, S. Casner, R. Frederick, and V. Jacobson, "RTP: ATransport Protocol for Real-Time Applications", RFC 1889.3 ITU Recommendation G.711 (11/88) Pulse code modulation (PCM) ofvoice frequencies.4 ITU Recommendation G.726 (12/90) 40, 32, 24, 16 kbit/s AdaptiveDifferential Pulse Code Modulation (ADPCM).5 ITU Recommendation G.727 (12/90) 5-, 4-, 3and 2-bits sampleembedded adaptive differential pulse code modulation (ADPCM).6 ITU Recommendation G.728 (09/92) Coding of speech at 16 kbits/susing low-delay code excited linear prediction.7 ITU Recommendation G.722 (11/88) 7 kHz audio-coding within 64kbit/s. ZopfStandards Track July, 20026 RTP Payload for Comfort NoiseJanuary 2002 8 Bradner, S., "