AssemblyAI

AI Voice

v0.4.0

Last updated Jun 23, 2026

n8n node for AssemblyAI speech-to-text transcription models.

Included Nodes

AssemblyAI

This is an n8n community node. It lets you use AssemblyAI in your n8n workflows.

AssemblyAI develops industry-leading Speech AI models for transcription and audio understanding, accessible through their API. It provides features like speaker diarization, sentiment analysis, entity detection, PII redaction, prompting (Universal-3 Pro), Medical Mode, and LLM Gateway capabilities for processing transcripts with AI.

Resources

• n8n community nodes documentation
• AssemblyAI API reference
• AssemblyAI API documentation

– AssemblyAI Speech-to-Text guide
– AssemblyAI LLM Gateway guide

Quick links

• Installation
• Credentials
• Operations
• Development
• Support & Feedback

Installation

Follow the installation guide in the n8n community nodes documentation.

Credentials

To use this node, you need an AssemblyAI API key.

Prerequisites

1. Sign up for a free account at AssemblyAI and receive $50 in free credits.
2. Navigate to your API Keys page in the dashboard.

Setup in n8n

1. In n8n, go to Credentials → New
2. Search for AssemblyAI API and select it
3. Enter your API key from the AssemblyAI dashboard
4. Choose your Data Region — US (Default) or EU (Data Residency). The EU region routes every request to api.eu.assemblyai.com / llm-gateway.eu.assemblyai.com, keeping audio and transcription data within the European Union.
5. Click Save

Learn more about API keys in the AssemblyAI documentation.

Operations

File (API Reference)

• Upload: Upload a media file to AssemblyAI for transcription

Transcript (API Reference)

• Create: Start a new transcription job with support for:

– Speech models: set speechmodels (priority list, e.g. universal-3-5-pro,universal-3-pro,universal-2) to choose a model. The API routes per language and falls back through the list. Available models: Universal, Universal-2, Universal-3 Pro, Universal-3.5 Pro. speechmodels is now the only way to select a model — the legacy singular speech_model field was removed in v0.4.0 (the API rejected pro models on it). Leave empty to let the API choose.
– Prompting (Universal-3 Pro / Universal-3.5 Pro): pass a prompt (up to 1500 words) to steer transcription style and accuracy
– Temperature (Universal-3 Pro / Universal-3.5 Pro): sampling temperature 0.0–1.0
– Medical Mode: set domain to medical-v1 for specialised medical-terminology accuracy
– Remove Audio Tags (Universal-3 Pro): strip inline annotations like [laughter], [music], and speaker cues
– Language detection with expectedlanguages, fallbacklanguage, codeswitching boolean (Universal-2), codeswitchingconfidencethreshold
– Code-switching transcription via the top-level language_codes field
– Speaker diarization with speakersexpected and speakeroptions (min/max speakers)
– PII redaction with redactpiireturnunredacted, overrideaudioredactionmethod (silence), and Redact Static Entities (custom label → terms map for literal find-and-replace)
– Keyterm prompting to boost recognition of domain-specific terminology
– Profanity filtering, sentiment analysis, entity detection, content safety, IAB categories
– Speech Understanding at create time: translation, speaker identification, custom formatting
– autochapters, summarization, summarymodel, summary_type are still available but deprecated — use the LLM Gateway resource instead.

• Get: Retrieve a transcription by ID
• Delete: Delete a transcription
• List: List all your transcriptions, filtered by status, date, ID range, or throttled_only
• Get Sentences: Get transcript broken into sentences
• Get Paragraphs: Get transcript broken into paragraphs
• Get Subtitles: Export subtitles in SRT or VTT format
• Get Redacted Audio: Get redacted audio URL
• Word Search: Search for specific words in the transcript

LLM Gateway (API Reference)

• Chat Completion: Send chat completion requests to LLMs (Claude, GPT, Gemini, Qwen, Kimi). See available models. Supports:

– Transcript injection: pass transcript_id to substitute {{ transcript }} in your prompt with the transcript’s text
– Tool calling: provide a JSON tools array; force a specific function via Tool Choice: Specific Function and a function name; multi-turn round-trips via the Tool message role + toolcallid
– JSON repair post-processing: opt-in toggle that sends postprocessingsteps: [{ type: "json-repair" }] to fix malformed JSON in tool-call arguments or structured outputs

• Speech Understanding: Process speech understanding tasks on existing transcripts:

– Translation: Translate transcripts into multiple languages
– Speaker Identification: Identify speakers by name or role
– Custom Formatting: Apply custom formatting to dates, phone numbers, and emails

Development

To work on this node locally, clone the repository and install dependencies:

git clone https://github.com/gsharp-aai/n8n-nodes-assemblyai.git
cd n8n-nodes-assemblyai
npm install

Build and run with linting:

npm run dev:fresh

The node will be available in n8n at http://localhost:5678

Support & Feedback

Need help or have feedback? Reach out to support@assemblyai.com.

Stay up to date with the latest features and improvements:

• AssemblyAI Product Changelog
• AssemblyAI Product Roadmap

n8n

n8n is a fair-code licensed workflow automation platform.

License

MIT