TTS Skill — Text-to-Speech & Animated Video Circles

Tool: tts

Convert text to speech. Video circles show animated robot avatar with audio waveform.

Providers

• Primary: OpenAI TTS (tts-1-hd) — most natural voice (needs OPENAI_API_KEY)
• Free: edge-tts Multilingual Neural — good quality, no API key needed

Parameters

• text (required) — text to speak
• voice (optional):
  • OpenAI: onyx (deep male), echo, fable, alloy, nova, shimmer
  • edge-tts multilingual (best free): brian (deep male, default), andrew (warm male), ava (female), emma (female)
  • edge-tts standard: ru-male, ru-female, en-male, en-female, uk-male, uk-female
• format (optional) — video_note (animated circle, default) or audio (voice message)

Rules

1. Default format is video_note — animated video with audio waveform on robot face
2. For long texts (>800 chars), text will be truncated to fit 60s video limit
3. Default voice is brian (Multilingual, natural sounding) — use it for most responses
4. The media is sent automatically after generation — just call the tool
5. Don't repeat the entire text in your message — just confirm it was generated
6. If user says "ответь кружочком" — summarize your response in a short text and send as video_note
7. When user sends a video circle, respond with a circle too (auto-hint is added)
8. Keep circle responses SHORT (2-4 sentences) — long text = long boring video