Support for Gemini tts and other audio model

[davit312]

I am trying to synthesize audio from text using gemin tts model

fabric -m gemini-2.5-flash-preview-tts \
    'Read this in friendly tone: How is your day going?' -o result.wav

and got this response

googleapi: Error 400: The requested combination of response modalities (TEXT) is not supported by the model. models/gemini-2.5-flash-preview-tts accepts the following combination of response modalities:
* AUDIO

Would be great if fabric supports audio responses.

[ksylvan]

I created this issue for your request #1607

[ksylvan]

@davit312 This feature is now in the latest Fabric as of v1.4.267

See https://github.com/danielmiessler/Fabric/blob/main/CHANGELOG.md#pr-1650-by-ksylvan-update-gemini-plugin-to-new-sdk-with-tts-support

Also keep an eye on #1651

[ksylvan]

@davit312 The shell completions are also updated.

fabric --help | grep -i tts
      --voice=                      TTS voice name for supported models (e.g., Kore, Charon, Puck) (default: Kore)
      --list-gemini-voices          List all available Gemini TTS voices

And hitting [TAB] after putting in --voice:

And here is the list of voices:

fabric --list-gemini-voices
Available Gemini Text-to-Speech voices:

Clear & Professional:
  Charon          - Informative and clear
  Rasalgethi      - Informative and professional
  Iapetus         - Clear and articulate
  Erinome         - Clear and precise

Natural & Expressive:
  Aoede           - Breezy and natural
  Pulcherrima     - Forward and expressive
  Lyra            - Melodic and expressive

Other Voices:
  Vindemiatrix    - Gentle and kind
  Achernar        - Soft and gentle
  Enceladus       - Breathy and soft
  Algieba         - Smooth and pleasant
  Vega            - Smooth and flowing
  Algenib         - Gravelly texture
  Callirrhoe      - Easy-going and relaxed
  Despina         - Calm and serene
  Gacrux          - Mature and experienced
  Fenrir          - Excitable and dynamic
  Zubenelgenubi   - Casual and conversational
  Sadaltager      - Experimental voice with a calm and neutral tone
  Umbriel         - Experimental voice with a deep and resonant tone

Firm & Confident:
  Kore            - Firm and confident (default)
  Orus            - Firm and decisive
  Alnilam         - Firm and strong

Bright & Cheerful:
  Puck            - Upbeat and energetic
  Laomedeia       - Upbeat and lively
  Zephyr          - Bright and cheerful
  Autonoe         - Bright and optimistic
  Leda            - Youthful and energetic
  Sadachbia       - Lively and animated

Warm & Welcoming:
  Sulafat         - Warm and welcoming
  Capella         - Warm and approachable
  Achird          - Friendly and approachable
  Schedar         - Experimental voice with a warm and engaging tone

Use --voice <voice_name> to select a specific voice.
Example: fabric --voice Charon -m gemini-2.5-flash-preview-tts -o output.wav "Hello world"

There's documentation here: https://github.com/danielmiessler/Fabric/blob/main/docs/Gemini-TTS.md

And there, you can find the information about adding a stanza to your Fabric config file to choose a default voice.

[davit312]

@ksylvan Thank you very much