Compare VoIP codecs by workflow, network, and interoperability.
Compare G.711, G.729 and Opus across bitrate, overhead, complexity, resilience, interoperability, licensing, transcoding, and business use.
Quick answer
A VoIP codec encodes and decodes audio. G.711 is a widely interoperable narrowband telephony baseline with a 64 kbit/s payload before packet overhead. G.729 uses a lower encoded bitrate and can involve licensing or implementation considerations. Opus supports a broad range of bitrates and audio bandwidths with packet-loss resilience, especially for modern application and WebRTC use. The correct choice depends on every endpoint and provider in the path.
- Page type
- Technical guide
- Evidence owner
- TalkChief Voice Engineering
- Content status
- Reviewed
- Last reviewed
Codec selection crosses the complete media path
Follow the operational path in reading order. Each stage remains visible when motion is reduced.
- 01
Capture and audio goal
Speech, music, narrowband phone, wideband app, and recording goals differ.
- 02
Encode and packetize
Codec, bitrate, packet interval, headers, encryption, and tunneling determine load.
- 03
Network behavior
Capacity, delay, jitter, loss, buffers, and concealment influence the result.
- 04
Interoperate or transcode
Endpoints, SBCs, providers, and destinations must share a codec or translate it.
- 05
Decode and assess
The recipient hears the combined effect of every media segment.
G.711, G.729, and Opus solve different constraints
G.711 is common in telephone networks and provides straightforward interoperability at a comparatively high payload rate. G.729 was designed for lower-bit-rate speech and may help on constrained links, but requires compatible implementations and review of current licensing or commercial conditions. Opus adapts across narrowband speech through full-band audio and is widely used in modern real-time applications.
Payload bitrate is not the complete bandwidth figure. Add IP, transport, RTP, link, encryption, tunnel, and packetization overhead in both directions, then include concurrency and headroom. Shorter packet intervals can reduce serialization and loss impact while increasing packets and header overhead.
Avoid unnecessary transcoding
When adjacent systems do not share a codec, a media function may decode and re-encode the audio. Transcoding adds processing, can add delay, may reduce quality, and creates another capacity and failure dependency. Multiple transcoding stages can compound the effect.
Choose an endpoint and provider codec policy that supports the required routes, recording, DTMF, conferencing, mobile networks, browser media, and destination interconnection. Confirm the negotiated codec in a real call rather than relying only on configuration preference.
Codec selection checklist
Test representative endpoints, routes, networks, and failure conditions.
Supported codecs and preference order at every endpoint, service edge, provider, and destination
Audio bandwidth and speech-quality requirement
Payload plus packet, encryption, tunnel, and link overhead
Packet interval, jitter buffer, loss concealment, FEC, and DTX behavior
CPU, battery, mobile network, browser, conference, recording, and transcoding impact
Licensing, commercial, regulatory, and interoperability constraints
Sources and review dates
These sources support the definitions and context on this page. Regulator material does not by itself prove that TalkChief holds a particular local permit, licence, or approval.
- ITU-T G.711 audio codingReviewed
- ITU-T G.729 speech codingReviewed
- IETF RFC 6716: Opus codecReviewed
Frequently asked questions
Which codec is best for VoIP?
There is no universal best codec. Choose based on endpoint and provider support, audio requirement, network conditions, overhead, resilience, transcoding, CPU, licensing, and the routes the business must use.
Does G.711 use only 64 kbit/s on the network?
64 kbit/s is the encoded payload rate. Actual network use is higher after RTP, transport, IP, link, encryption, tunnel, and packetization overhead.
Why can a wideband call become narrowband?
If any endpoint, provider, public-network segment, conference, recording, or transcoder supports only a narrower format, the end-to-end call can be limited or transcoded.