Voice and audio
ChatHaiku can read replies aloud in one of eleven voices, and can take dictation so you can speak instead of type.
Reading replies aloud
There are two ways to hear a reply. The speaker button under any reply reads that one on demand. The Speak responses switch in Settings → Voice reads every reply automatically as it arrives.
The speaker icon reflects state: a spinner while audio is being prepared, and animated waves while it plays. Press it again to stop. Starting a new reply, switching conversations, or navigating away also stops playback.
Available voices
| Accent | Voices |
|---|---|
| American — female | Heart (default), Bella, Nicole, Sarah, Sky |
| American — male | Adam, Michael |
| British — female | Emma, Isabella |
| British — male | George, Lewis |
Speech is generated by Kokoro, running on Rootcomputer's hardware rather than your browser's built-in voices. That is why the voices sound the same on every device — and why speech needs a working connection and can be briefly unavailable if that service is down.
Because speech is generated server-side, the text of a reply is sent to Rootcomputer's speech service when you play it. Only the reply text is sent — not your messages, and not the rest of the conversation.
Dictation
The microphone button beside the composer starts dictation. Speak, and your words appear in the input box; press the button again to stop. Recognition also ends on its own after a pause.
While recording, a live waveform slides in above the composer so you can see that your microphone is actually being picked up.
Dictation uses your browser's speech recognition rather than Rootcomputer's servers. Where that processing happens depends on your browser and platform — most send audio to the browser vendor's own service. Nothing about dictation passes through ChatHaiku until you send the message.
Turning the microphone off
Settings → Privacy has a Microphone switch, on by default. Turning it off means dictation cannot start: the button stays where it is and explains that access is disabled, rather than disappearing or silently doing nothing. Switching it off also stops any dictation already in progress.
Not every browser supports speech recognition. Rather than showing a control that would fail, ChatHaiku hides the button where it is unavailable. Chrome and Edge have the broadest support; Firefox generally does not. You will also be asked for microphone permission the first time, and denying it prevents dictation until you change it in your browser's site settings.
On Android, the waveform may not appear even though dictation works normally. The system speech service takes exclusive control of the microphone there, so the visualiser is skipped rather than fighting it for access.
If audio does not play
- Nothing happens on the first press. Browsers block audio until you have interacted with the page. Press again.
- The spinner never resolves. The speech service may be down — check the status page.
- Playback cuts off. Sending a new message stops the previous reply's audio by design.
- Silent on iOS. Check the hardware mute switch, which silences web audio independently of volume.