A high-performance transcription skill using Deepgram Nova-3 for lightning-fast, accurate audio-to-text conversion with speaker diarization.
Install & Download
1. ClawHub CLI
The fastest way to install a skill directly from the registry.
npx clawhub@latest install deepgram-transcribe
2. Manual Installation
Copy the skill folder to one of these locations
~/.openclaw/skills/ <project>/skills/ Priority: Workspace > Local > Bundled
3. Prompt Installation
Copy this prompt to OpenClaw to install it automatically.
Help me install deepgram-transcribe using Clawhub. If Clawhub is not installed, install it first (npm i -g clawhub).
Prefer to download?
Get the raw skill files in a ZIP archive.
What is Deepgram Transcribe?
The Deepgram Transcribe skill brings industrial-grade speech-to-text capabilities to your AI agent. By utilizing the Deepgram Nova-3 model, this skill achieves a significantly lower word error rate (5.26%) and operates 40x faster for batch processing compared to traditional Whisper models. It is an essential addition to your collection of Openclaw Skills when precision and speed are critical for processing audio data.
This skill is designed to handle a wide variety of audio formats including mp3, wav, m4a, and flac. It doesn't just transcribe; it intelligently formats numbers, dates, and punctuation while providing built-in speaker diarization to identify who said what, making it far more capable than standard transcription tools.
Deepgram Transcribe Use Cases
- Transcribing long-form podcasts and interviews with high accuracy.
- Converting recorded meetings into structured text for documentation.
- Processing voice memos and recordings into searchable text formats.
- Automating transcription for bulk media libraries at low latency.
- Generating JSON-formatted metadata for audio files including timestamps and confidence scores.
How Deepgram Transcribe Works
- The user initiates a transcription command providing an audio file path.
- The skill validates the presence of a Deepgram API key in the secrets directory or environment.
- The audio file is transmitted to the Deepgram API using the specified model (Nova-3 by default).
- The API processes the audio, applying speaker diarization and smart formatting.
- The skill returns the formatted transcript as a text file or a detailed JSON object.
- If the primary API is unavailable, it gracefully falls back to the OpenAI Whisper skill.
Deepgram Transcribe Setup
To use this within your suite of Openclaw Skills, you must first store your Deepgram API key. You can do this by creating a secrets file:
echo "YOUR_DEEPGRAM_API_KEY" > ~/.openclaw/secrets/deepgram-api-key.txt
Alternatively, you can set an environment variable in your shell configuration:
export DEEPGRAM_API_KEY="your-key-here"
Deepgram Transcribe Data Schema & Taxonomy
The skill produces transcription data in multiple formats based on the user's flags. Below is the organization of the output data:
| Feature | Output Detail |
|---|---|
| Default Output | A plain text file (.txt) containing the full transcription. |
| JSON Output | A structured .json file containing timestamps, speaker labels, and confidence metrics. |
| Supported Formats | mp3, wav, m4a, ogg, flac, webm, aiff. |
| Model Options | nova-3 (default), nova-2 (budget), whisper-large (parity). |
Deepgram Transcribe Advanced Features
- Speaker Diarization: Automatically distinguishes between different speakers in a conversation.
- Smart Formatting: Automatically formats dates, currency, and phone numbers for better readability.
- Multi-Language Support: Transcribe audio in various languages using the
--languageflag. - Automated Fallback: Integrated logic to switch to OpenAI Whisper if Deepgram credentials are not found.
- Batch Processing: Optimized for 40x faster throughput compared to standard Whisper implementations.
SKILL.md
Loading
Related Openclaw Skills
Philips Hue Thinking Indicator
A physical status indicator for AI agents that uses Philips Hue lights to signal thinking and completion states.
Xiaohongshu Auto-Publish
Automate the publication of image-rich posts to Xiaohongshu via browser-based CDP control.
Docs Improver
A professional suite for auditing, generating, and synchronizing high-quality technical documentation for any software project.
Code Analyzer
A professional-grade deep repository analysis engine that uncovers architecture styles, execution flows, and Domain-Driven Design patterns across 20+ languages.
Host Security Audit
A professional-grade security auditing tool for OpenClaw host machines that identifies vulnerabilities and system health risks.
Meeting Transcripts (Fireflies.ai)
Automatically capture, summarize, and organize meeting transcripts from Fireflies.ai into structured Markdown files.
Featured*
