Audio codecs negotiation and support

Overview

An audio codec defines how voice audio is compressed, encoded, transmitted, and decoded during a call. In SIP calls, codecs determine the audio quality, bandwidth usage, and compatibility between endpoints.

Different codecs trade off quality, latency, and network efficiency. For example, uncompressed, or lightly compressed codecs offer higher quality but use more bandwidth, while highly compressed codecs conserve bandwidth at the cost of audio fidelity.

Choosing the right codec matters because both sides must support the same codec to exchange audio. If both sides can't negotiate a common codec, the call might connect, but audio exchange fails between endpoints or requires an external service to transcode the audio.

SDP offer and answer

As part of the SIP handshake, codecs are negotiated by exchanging Session Description Protocol (SDP) messages. SDP  is a text-based format used to describe media capabilities including codecs, ports, and other associated properties, to negotiate communication between endpoints.

SDP offer

The SDP offer lists all audio codecs the caller supports, in order of preference. The caller frequently includes an SDP offer in the initial INVITE. This message lets the callee know the codecs the caller supports and allows the callee to select the best codec to use for the call.

SDP offer example

The following example is a simplified SDP offer in an INVITE:

m=audio 49170 RTP/AVP 0 8 96
a=rtpmap:0 PCMU/8000
a=rtpmap:8 PCMA/8000
a=rtpmap:96 opus/48000/2

What it means:

• The caller is offering PCMU (G.711 µ-law), PCMA (G.711 A-law), and Opus.
• The order indicates preference.

SDP answer

The callee selects one or more codecs from the offer list, in order of preference, and returns an SDP answer in the 200 OK response. This message lets the caller know the codec the callee selected and allows the caller to confirm the codec selection.

The following is an example SDP answer in a 200 OK response:

m=audio 53000 RTP/AVP 0 8
a=rtpmap:0 PCMU/8000
a=rtpmap:8 PCMA/8000

What it means:

• The callee chose PCMU/8000 as the codec.
• Both sides now send and receive audio using the PCMU/8000 codec.

When does media start?

Codec negotiation completes when the caller sends an ACK for the 200 OK. RTP audio typically starts flowing immediately after the ACK is received. For handshake details, see Caller acknowledges the final response.

Supported audio codecs

LiveKit supports the audio codecs in the following sections.

PCMU (G.711 µ-law)

A lightly compressed, narrowband codec operating at 8 kHz. It offers high compatibility and low latency but uses more bandwidth (~64 kbps plus overhead). PCMU is widely used in North America and is a baseline codec for PSTN interoperability.

PCMA (G.711 A-law)

Similar to PCMU in quality, frequency, and bandwidth, but uses A-law companding.

G.722

A wideband codec that operates at 16 kHz, providing higher audio quality (HD voice) while using a similar bitrate to G.711.

AMR-WB

Adaptive Multi-Rate Wideband (AMR-WB) is a wideband codec that operates at 16 kHz, providing HD voice quality. It isn't enabled by default. To use it, add it to the list of extra codecs for a call. To learn more, see Configuring codecs.

Configuring codecs

By default, LiveKit SIP offers a fixed set of codecs in the SDP it sends. You can customize this set per call to enable additional codecs or restrict the call to a specific list. Codecs are configured through the media field (SIPMediaConfig) on inbound trunks, outbound trunks, dispatch rules, and CreateSIPParticipant requests.

Default codecs

LiveKit SIP enables a set of SIP codecs by default. These are distinct from the codecs used inside a LiveKit room. The default SIP codecs are PCMU, PCMA, and G722.

Enable extra codecs

To enable a codec that isn't on by default, add it to the codecs list in the media config. The codec name must exactly match the value used in SDP. For example, to enable AMR-WB in addition to the default codecs:

{
  "media": {
    "codecs": [{ "name": "AMR-WB" }]
  }
}

The full list of supported codec names is as follows:

• PCMU
• PCMA
• G722
• AMR-WB

LiveKit doesn't enable every supported codec by default because each codec added to the SDP increases the size of the INVITE packet. When a call uses UDP transport, large packets can fragment and get lost on some networks. Enabling codecs explicitly lets you control this tradeoff.

Restrict to specific codecs

The codecs list only adds codecs on top of the defaults, so there's no direct way to remove a default codec. To send only a specific set of codecs, set only_listed_codecs to true. When set, LiveKit SIP clears the default codecs for the call and uses only the codecs in the codecs list.

You have two options depending on your transport and requirements:

• Add codecs to defaults (TCP/TLS only): Leave only_listed_codecs unset and add extra codecs to the codecs list. Use this on TCP/TLS connections, which provide reliable delivery and automatically handle data segmentation.
• Use only specific codecs (UDP, or when avoiding defaults): Set only_listed_codecs to true and specify the complete codec list in codecs. The call uses only those codecs. Use this to control packet size when the call uses UDP transport or when you need to exclude a default codec that a peer doesn't support.

The following example sends only PCMU and AMR-WB:

{
  "media": {
    "onlyListedCodecs": true,
    "codecs": [{ "name": "PCMU" }, { "name": "AMR-WB" }]
  }
}

Codec sample rate

Each SIPCodec entry also accepts a rate parameter that sets the sample rate. Each currently supported codec uses a single sample rate, so if rate is omitted, the default rate for the codec is used.

For reference, the supported codecs and their sample rates are as follows:

Additional resources

The following resources provide additional details about the topics covered in this guide.