Audio replies (text-to-speech voice output)

By default the AI replies in text. You can turn ON audio replies on the Voice page; from then on, replies are sent as audio messages on WhatsApp and Telegram (channels that support audio). The website widget always uses text.

Picking a voice:
- 1 language only in your settings: native voices for that language are listed. The AI picks one of those.
- 2+ languages: only multilingual voices are listed (so the same voice handles every language).
- Cloned voice: if you cloned one, it appears at the top of the list.

Audio toggle: when OFF, audio replies are disabled and the AI defaults back to text. When ON, the AI decides per message: short replies stay text, fuller replies are sent as voice. You can override the auto-decision per contact in the Contact rules section.

Costs: TTS adds a small cost per character; check the Plans & Pricing page for current limits per plan.

Tips:
- For sales and personal-feeling channels, audio noticeably lifts response rates.
- Keep emojis OFF when audio is ON: they read awkwardly.
- Test the voice with the Preview button before going live.