Audio Settings

Audio Settings & Voice Activation

WebSpeak3 currently transmits voice using voice activation (it starts sending audio automatically once you speak loud enough) — there is no push-to-talk yet. This page covers how to tune that, and the other audio options available.

Playback (Options → Playback)

• Ausgabegerät (Output device) — pick which speaker/headset WebSpeak3 plays audio through. Works even in browsers that don't support AudioContext.setSinkId natively.
• Voice volume — overall playback volume, shown in dB.
• Play test tone — plays a short tone through the selected output device, so you can check it's the right one before joining a server.

Recording (Options → Capture)

• Mikrofon (Input device) — pick which microphone to use.
• Voice activation threshold — the slider under the mic level bar. Speak normally and watch the live level bar; set the threshold just above your room's background noise level so WebSpeak3 doesn't key up on silence, but does key up as soon as you talk.
• Hangover delay — how many seconds WebSpeak3 keeps transmitting after your volume drops back below the threshold, so short pauses mid-sentence don't cut you off. Increase it if you're getting clipped; decrease it if your channel hears too much trailing silence after you stop talking.
• Test button — lets you hear your own mic through your selected output device (with the level bar active) so you can tune the threshold without actually connecting to a server.
• Echo cancellation / Noise suppression / Automatic gain control — standard browser-level audio processing toggles. Turn noise suppression off if it's cutting into your voice or reacting strangely to background sound; turn echo cancellation off only if you're using headphones and it's causing audio artifacts.

Speaker + microphone setups (no headset)

If you use speakers and an open microphone instead of a headset, echo can be harder for the browser to fully cancel — the mic picks up your own speaker output acoustically. A few things to try, in order:

1. Keep echo cancellation on (it's on by default) — turning it off will make things worse in this setup.
2. Try toggling automatic gain control off. AGC re-amplifies your mic signal after echo cancellation has already run, which can bring back faint echo that the canceller mostly removed — some speaker+mic setups sound noticeably cleaner with AGC off, at the cost of your own volume being less automatically leveled.
3. Lower your speaker volume, and/or increase the physical distance between your mic and speakers, if possible — this reduces how much echo the AEC engine has to cancel in the first place, which any amount of software-side processing benefits from.
4. If it's still not clean enough, a headset remains the reliable fix — the browser's built-in echo cancellation (used by all major browsers, including WebSpeak3) is primarily tuned for headset/laptop-mic scenarios. This is a known limitation of browser-based voice chat in general, not something specific to WebSpeak3; see issue #2 for the current state of investigating a stronger, browser-independent fix.

Permissions

Your browser will ask for microphone permission the first time you enable the mic. If you don't see a prompt, or audio never activates:

• Check your browser's site settings for the WebSpeak3 URL — the mic permission may have been previously denied and needs to be reset there.
• Microphone access requires a secure context. It works on localhost (and on the official hosted client/demo, which are HTTPS). If you're self-hosting and accessing WebSpeak3 over plain HTTP from another device, the browser will block mic access — put it behind an HTTPS reverse proxy for that.

Known limitations

Safari and Mobile Safari's audio/microphone behavior is still marked experimental — see the beta roadmap for the current status. If something doesn't work as expected there, a bug report with your Safari version and OS is genuinely useful.