Skip to content
AI Dialer

Glossary · Voice technology

What is Codec?

  • Also called: G.711, Opus, G.729

Reviewed by Sujan ThapaliyaLast updated

Definition

A codec encodes and compresses voice for transmission over a network. G.711 is uncompressed and highest quality at about 64 kbps; G.729 compresses to roughly 8 kbps with audible loss; Opus adapts its bitrate and handles poor networks best.
The three STIR/SHAKEN attestation levels, which underpin how codec behaves in practice: A asserts both that the carrier knows the customer and that the customer may use the calling number, B asserts the customer only, and C asserts neither. A shared number pool can reach B at best, which is why number ownership keeps recurring in these definitions.
Attestation is the backdrop to most caller-ID and deliverability terms, codec included.

In practice

How codec actually works

Transcoding between codecs on a call path adds latency and degrades quality. Matching codecs end to end is worth more than picking the theoretically best one.

For AI voice agents the codec matters more than for humans: compression artefacts that a person's ear ignores measurably reduce speech-recognition accuracy.

Codec negotiation is also a common source of one-way audio between systems that each support several. Both ends advertise a list, pick the highest common option, and if an intermediate device rewrites that list badly the call connects with no agreed codec in one direction. It presents as a network fault and is a signalling one.

Each transcode between codecs costs quality, so two conversions sound worse than one poor codec.
CodecBandwidth per directionUse when
G.711~64 kbps before overheadBandwidth is plentiful; best quality
Opus6–128 kbps, adaptiveThe network is variable — it adjusts
G.729~8 kbpsBandwidth is genuinely constrained; quality cost is audible

Worked example

Twenty concurrent G.711 calls need roughly 2 Mbps of upstream before overhead. The same twenty on Opus can run in well under half that, which is why the codec choice is a network decision as much as an audio one.

Codec: common questions

Which codec should I use?
G.711 where bandwidth is plentiful, because it is uncompressed and sounds it. Opus where the network is variable, because it adapts. G.729 only where bandwidth is genuinely constrained, since the quality cost is audible.
How much bandwidth does a call need?
G.711 uses about 64 kbps per direction before overhead, so roughly 80–100 kbps in practice. Opus varies with conditions and can run well under that. Multiply by concurrent calls, then leave headroom for everything else on the connection.
Does codec choice affect how calls sound to the recipient?
Yes, and transcoding makes it worse. Every conversion between codecs along the path costs quality, which is why a call that sounds poor is often being converted twice rather than encoded badly once.

Sources

  1. Combating Spoofed Robocalls with Caller ID Authentication — Federal Communications CommissionThe STIR/SHAKEN framework, the attestation levels carriers sign calls with, and the mandate requiring providers to authenticate caller ID.
  2. 47 U.S.C. § 227 — Restrictions on the use of telephone equipment — Cornell Legal Information InstituteThe Telephone Consumer Protection Act itself — the consent requirements, calling-hours limits, and private right of action.
  3. ITU-T Recommendation E.164 — The international public telecommunication numbering plan — International Telecommunication UnionThe international number format, the 15-digit maximum, and how country codes and national numbers compose.

See codec in the product

SIP trunking carries your phone calls over the internet to and from your existing PBX, replacing physical ISDN or PRI lines. You keep your phone system and your numbers; you drop the copper, the per-channel rental, and most of the long-distance bill.

Explore sip trunkingGet early access