n8n-nodes-speech-to-text-cloud

This is an n8n community node that integrates with Online Speech To Text Cloud, a premium transcription service. Transcribe audio/video files, apply text transformations (summarize, translate, diarize), and retrieve structured results directly in your n8n workflows.

Installation | Operations | Credentials | Compatibility | Usage | Troubleshooting | Resources

Installation

Follow the installation guide in the n8n community nodes documentation.

Operations

| Resource | Operation | Description | |----------|-----------|-------------| | Job | Transcribe | Start a new transcription job from an audio file | | Job | Get Result | Fetch the transcript, status, or transformation output | | Job | Apply Task | Run an LLM task (summarize, translate, diarize, etc.) on a completed job |

Credentials

1. Sign up at Online Speech To Text Cloud and generate an API key starting with stt_live_...
2. In n8n, create a new Online Speech To Text Cloud API credential and paste your key.
3. The node automatically attaches the key via the x-api-key header.

Compatibility

• n8n version: ≥ 1.0.0 (tested with 1.20+)
• Node.js: ≥ 18.17.0
• Works in n8n Cloud, self-hosted, and community editions.

Usage

Transcribing Audio

1. Download the file: Use an HTTP Request node to fetch your audio from a URL. Set Response Format to File.
2. Configure Transcribe Node:
   • Audio File: Enter the name of the binary property from the previous HTTP Request step (usually data)
   • Language Code (Optional): Provide a 2-letter code (e.g., en, de, fr). Leave as yyy for auto-detection.
3. Send: The node will send your audio file as multipart/form-data to the API.

Example workflow:

HTTP Request (download audio) ↓ Transcribe (this node) ↓ Get Result (poll for completion)

Getting Results & Polling

The Get Result operation returns the current status and transcript. Since n8n declarative nodes don't include built-in polling loops, wrap it in a simple workflow loop:

1. Add a Wait node (set to Resume or Timeout).
2. Connect: Get Result → Wait → If (check status === 'finished')
3. Loop until finished, then proceed.

Example polling pattern:

Get Result ↓ If (status = 'finished') ├─ YES → Extract transcript data └─ NO → Wait (10 sec) → Loop back

Applying LLM Tasks

Use Apply Task to run transformations on an existing completed job:

• Task Type: summarize, translate, diarize, compliance
• Language (for translations): Target language code (e.g., es, de) or yyy for other tasks
• Job ID: The ID from your transcription job

Troubleshooting

"Missing boundary in multipart" error

• Ensure the binary property name is correct (usually data from HTTP Request)
• Verify the previous node is returning file data (not text)

"Field required" on audio_file

• Check that the HTTP Request node has Response Format set to File
• Confirm the property name matches exactly what the upstream node outputs

Timeout waiting for transcription

• Large files may take longer; increase the Wait node timeout
• Check job status via Get Result to see if processing failed or is still in progress

API key not working

• Verify your API key starts with stt_live_
• Check that the credential is properly saved in n8n
• Ensure you're using the correct credential in the node configuration

Resources

• n8n community nodes documentation
• Online Speech To Text Cloud API Docs
• Pricing & Plans