Voice Input & Playback
Chat supports voice on both sides: voice questions (record → auto-transcribe → send) and voice answers (TTS playback). Each side works independently — for example "voice in / text out" or "text in / voice out".
Voice Input (Record a Question)
- Click the mic icon in the input bar ("Click to start talking");
- Click again to start recording; the icon area shows "Talking (N s)";
- Click the icon once more to finish — the recording is sent automatically, no extra send click;
- The audio is transcribed by the system's ASR service and sent as the question.
Notes:
- The ASR service is configured globally by the admin (only one active at a time); users cannot switch it;
- Max duration and file size follow the system configuration (e.g. 60 seconds); overlong recordings are cut off;
- The transcription becomes the question text and cannot be edited before sending — ask again by text if the recognition is wrong.
Voice Playback (AI Replies)
Controlled in the character's settings:
| Setting | Options | Description |
|---|---|---|
| AI reply format | Auto / Text / Voice | "Auto" follows your input format |
| Auto-play voice replies | On / Off | Play automatically when the answer arrives |
| Voice | Dropdown | Server-side TTS: pick one; browser-side synthesis: not needed |
For voice answers, click Show text / Show audio to switch between them — text to check the content, audio to listen.
NOTE
TTS is also configured by the admin. Whether synthesis happens server-side or in the browser depends on deployment: browser-side needs no voice selection, and available voices may differ.
For Developers
Integrating new ASR/TTS platforms is development work — see ASR Integration and TTS Integration.
Previous: Character Settings · Next: Drawing