Asterisk codecs and transcoding
A configured codec list describes a policy. The codec negotiated on a particular call leg and any conversion Asterisk performs are runtime facts. Keep all three distinct when diagnosing audio or estimating host load.
Source-reviewed, 18 October 2026. Prepared from the cited documentation. Examples and performance have not been independently tested by Phone Guides. Scope: Asterisk 22/PJSIP audio calls; no measured call count or endpoint compatibility is claimed.
Read the policy first
An endpoint can restrict its allowed codecs. The introductory two-phone example uses the following policy fragment:
disallow=all
allow=ulaw,alaw
Those names select G.711 variants in Asterisk's configuration vocabulary. The actual endpoint must support the permitted codecs, and the required codec modules must be available. An allowed value in a file is not proof that a usable path was negotiated.
Use the branch-22 PJSIP reference for endpoint options. For broader codec planning, use the Phone Guides codec reference without treating planning figures as measurements of your call.
Inspect both legs
In a call between two endpoints, Asterisk can have separate media negotiations on the incoming and outgoing sides. A codec preference at one endpoint does not by itself tell you the final format on the other leg. Compare the offer/answer and actual channel information for both.
The upstream advanced codec negotiation guide describes the negotiation stages. It includes a historical development preface, so use it for the model and compare configuration details with the current branch reference and installed help. Do not copy its example defaults as a claim about every patch.
Distinguish conversion from pass-through
When legs use different audio formats and Asterisk handles the media, it needs a supported translation path. A module being available does not establish that every codec pair, recording format or bridge mode can use it. Direct media, recording, conferencing and application processing can change the work required.
| Question | Reader evidence |
|---|---|
| What codecs are configured? | Loaded endpoint policy |
| What did each side negotiate? | Offer/answer and channel media information |
| Can this build translate the formats? | Loaded codec modules and translation information |
| What load does the call create? | Measurements on the actual host and workload |
| Is the audio path usable? | Two-way audio and media-path checks |
Consult the installed CLI's help for format, codec and translation inspection commands. core show channel with an actual active channel name can help connect media observations to a particular call leg. Do not substitute a generic codec benchmark for call-path evidence.
Explain failed negotiation carefully
A call rejected during media negotiation may have no acceptable format under the configured policy. Read the response sequence and the application's log context before assigning a cause. Changing to an unrestricted codec list can obscure the mismatch and permit a format the rest of the system does not handle as intended.
If a call answers but audio is missing, also inspect advertised addresses and RTP. Codec agreement does not establish reachability. If speech works but digits fail, examine DTMF separately.
Measure before promising a host size
For sizing, record concurrent calls, codec mix, conversion, packetisation, recording/conferencing, encryption and logging. Keep host CPU/memory and the workload together with the result. A historical book's hardware table or a codec's nominal payload rate is not evidence for the number of calls your server can handle.
Preserve the previous endpoint policy, apply one justified change and inspect a new call. Record whether the intended format and behaviour actually changed. Return to Asterisk documentation.