n8n-nodes-groq

This is an n8n community node package that provides integration with the Groq API for Speech-to-Text functionality.

Features

- Speech to Text: Convert audio files to text using Groq's Whisper models
- Multiple input formats: Support for binary data and audio URLs
- Advanced models: Whisper Large V3, Whisper Large V3 Turbo, and Distil Whisper Large V3 EN
- Flexible response formats: Plain text, JSON, and detailed JSON with timestamps
- Advanced settings: Temperature control, language, prompts, and timestamp granularity

Installation

To use this community node, you need to install it in your n8n instance:

``bash
npm install n8n-nodes-groq
``

Configuration

1. Groq API Key: Get your API key from console.groq.com/keys
2. Credentials: Configure Groq API credentials in n8n with your API key

Usage

$3

The node allows you to convert audio to text using the following parameters:

- Audio Input:
- Binary data (from previous nodes)
- Audio file URL
- Model: Choose between Whisper Large V3 Turbo (fastest), Whisper Large V3 (high accuracy), or Distil Whisper Large V3 EN (optimized for English)
- Response Format: Text, JSON, or detailed JSON
- Language: Specify audio language (optional)
- Prompt: Text to guide model style (optional)
- Temperature: Control output randomness (0-1, optional)

$3

The node supports the following audio formats:
- WAV
- MP3
- M4A
- FLAC
- WEBM
- OGG
- OPUS

Example Usage

1. Use an "HTTP Request" or "Read Binary File" node to get an audio file
2. Connect to the Groq Speech to Text node
3. Configure the "Speech to Text" operation
4. Select "Binary Data" as audio input
5. Configure the model and other parameters as needed
6. Execute the workflow to get the transcription

Perfect for WhatsApp Integration

This node works seamlessly with WhatsApp voice messages:
1. Receive WhatsApp webhook with audio
2. Convert base64 audio to file using "Convert to File" node
3. Process with Groq Speech to Text
4. Send transcription back to WhatsApp

License

MIT

Support

For issues and suggestions, please open an issue in the project repository.