What is a voice codec?

Codec is short for coder-decoder, the software that squeezes your voice down for sending and rebuilds it on the other end. Here is the trade-off every codec is quietly making.

Compressing a voice for the trip

A codec, short for coder-decoder, is the software that converts your voice from raw sound into a compact stream of data for sending, and back into sound again on the other end. Without a codec, an uncompressed audio stream would use far more bandwidth than most connections can comfortably spare.

The trade-off every codec makes

Every codec balances the same three things against each other: sound quality, bandwidth used, and how much processing power it takes to compress and decompress in real time. A codec that sounds better usually costs more bandwidth, or more computation, or both.

G.711 is the oldest and simplest, barely compressing the audio at all, which makes it sound excellent but heavy on bandwidth. Opus, the modern default for most WebRTC calls, compresses far more aggressively while still sounding close to G.711 in good conditions, and it can adapt in real time as your connection quality changes mid-call.

Why a mismatch breaks a call

Both sides of a call have to agree on a shared codec before any audio can flow, which is exactly what gets negotiated inside the SDP exchange at the start of a call. If the two sides genuinely share no common codec, the call can connect completely and still produce total silence, one of the quieter and more confusing ways a call can fail.