Workflow for Gemini 2.5 Pro Preview TTS

Hello! Welcome!

Currently, callin.io does not appear to directly support audio generation using the gemini-2.5-pro-preview-tts model. While the model is listed within the Gemini Chat models in callin.io, you're unable to configure the responseModalities parameter necessary for an audio response.

To utilize the
gemini-2.5-pro-preview-tts model for speech synthesis, you'll need to perform a direct HTTP request to the Gemini API, ensuring the required parameters are correctly set according to the documentation you provided.

Set up an HTTP Request node in callin.io

• Method: POST
• URL:

 https://generativelanguage.googleapis.com/v1beta/models/gemini-2.5-pro-preview-tts:generateSpeech?key=YOUR_API_KEY 

• Headers:

Content-Type: application/json

• Body (Raw JSON):

{
  "text": "{{ $json.text }}",
  "audioConfig": {
    "speakingRate": 1.0,
    "voice": {
      "name": "en-US-Standard-B"
    }
  }
}

Please substitute {{ $json.text }} with the actual text output from your Agent node.

Decode and save the audio

The response will include a base64-encoded audio file. You can add a Function node with the following code:

return [{
  binary: {
    data: {
      data: Buffer.from($json.audio.audioData, 'base64'),
      mimeType: 'audio/mp3',
      fileName: 'output.mp3'
    }
  }
}];

Upload to Google Drive

Let me know! Cheers

I managed to get it working by implementing the following steps:

(One important detail is that the default output audio format is a .pcm file. This means you'll need to convert it to either WAV or MP3 for usability. If you're self-hosting callin.io, you can achieve this by installing ffmpeg into your Docker container. However, if you're using a cloud-based setup, you might need to utilize an external API service for the conversion.)

I tried your workflow. I can use Google Gemini TTS. Does Google Gemini require using ffmpeg to convert to a WAV file?

Thank you very much!

Yes, I believe Gemini only provides audio files in .pcm format, requiring conversion to .wav or .mp3 for usage. If you're self-hosting callin.io, I found this method to be the most straightforward.

This is fantastic! Just a heads-up: your API key is visible in the HTTP request.