Voice Chat

General definition

Voice Chat is live audio between users of the same app, carried over the internet rather than the phone network. The phrase usually implies a room model: a voice channel exists, people join it, talk, and leave, and the audio is mixed or forwarded so that everyone present hears everyone else. A voice call, by contrast, is point-to-point or invited: someone rings, someone answers, the session ends when they hang up. Both use the same VoIP technology underneath; the difference is the product shape.

Gaming popularised the room model, where a party channel stays open for hours and players drop in and out. Community platforms adopted it for stage-style rooms and audio spaces. In healthcare and financial services the same capability appears as audio-only consultations and advisor check-ins, where a voice channel attached to an existing secure chat thread is simpler and more private than a video appointment, and easier to audit than a phone call.

• Open mic transmits whenever voice activity detection hears speech; push-to-talk transmits only while a key is held, which cuts background noise and bandwidth in noisy or crowded rooms
• Opus (RFC 6716) is the standard codec: it covers narrowband speech to full-band music from about 6 kbit/s upward, adapts its bitrate on the fly and is mandatory in WebRTC
• Audio processing runs before encoding: echo cancellation stops your speaker output feeding back into your microphone, noise suppression removes fans and keyboards, and automatic gain control evens out loud and quiet speakers
• Group voice needs a server: an SFU forwards each speaker’s stream to the others, or a mixer combines them into one; either way clients only upload once

Voice is cheap on the wire, typically 20 to 40 kbit/s per speaker, so the quality problems are almost never bandwidth. They are latency, jitter and packet loss, and a mute button that works. A good voice chat implementation also carries presence and speaking indicators alongside the audio so users can see who is in the room and who is talking.

In the Ethora ecosystem

Ethora provides in-app voice as audio-only calls that a product can enable as an opt-in alongside, or instead of, video. Calls run on WebRTC through an SFU media server, support groups, and use the same SDK and the same signed-in identity as the chat rooms, so a voice session belongs to the conversation it started from rather than living in a separate tool. VoIP push notifications wake the app on iOS and Android when a call comes in, and the chat itself carries presence and typing state so users can see who is around before they speak. The voice and video call API page lists the calling features in the current SDK.

For regulated use the deployment model matters as much as the codec: the media server can run in managed cloud, in a dedicated server inside your own cloud account, or self-hosted, so audio for a patient or client conversation stays on infrastructure your organisation controls and can be governed by the same access controls and audit trail as the messages around it.