Faster Whisper
Local speech-to-text using faster-whisper — 4-6x faster than OpenAI Whisper with identical accuracy. GPU acceleration, SRT/VTT/TTML subtitles, speaker diarization, and chapter detection.
Install on your platform
We auto-selected Claude Code based on this skill’s supported platforms.
Run in terminal (recommended)
claude mcp add faster-whisper npx -- -y @trustedskills/faster-whisper
Or manually add to ~/.claude/settings.json
{
"mcpServers": {
"faster-whisper": {
"command": "npx",
"args": [
"-y",
"@trustedskills/faster-whisper"
]
}
}
}Requires Claude Code (claude CLI). Run claude --version to verify your install.
About This Skill
What it does
faster-whisper provides local speech-to-text transcription capabilities that are significantly faster – up to 6 times quicker – than the standard OpenAI Whisper model while maintaining comparable accuracy. It supports GPU acceleration for improved performance and offers features like SRT/VTT/TTML subtitle generation, speaker diarization (identifying different speakers), and automatic chapter detection within audio files. This allows for efficient transcription of long-form audio content without sacrificing quality.
When to use it
- Transcribing lengthy lectures or presentations where speed is crucial.
- Generating subtitles for videos in SRT, VTT, or TTML formats for accessibility or distribution.
- Analyzing meeting recordings and identifying different speakers involved in the conversation.
- Creating chapter markers within podcasts or audiobooks to improve navigation.
Key capabilities
- GPU acceleration for faster transcription speeds
- SRT/VTT/TTML subtitle generation
- Speaker diarization
- Chapter detection
Example prompts
- "Transcribe this audio file and generate an SRT subtitle file."
- "Identify the speakers in this meeting recording."
- "Create chapter markers for this podcast episode based on pauses and content changes."
Tips & gotchas
- A GPU is highly recommended to take full advantage of faster-whisper's speed improvements. Without a GPU, transcription will be slower.
- The accuracy of speaker diarization can depend on audio quality and the clarity of different speakers’ voices.
Requirements
TrustedSkills Verification
Unlike other registries that point to live repositories, TrustedSkills pins every skill to a verified commit hash. This protects you from malicious updates — what you install today is exactly what was reviewed and verified.
Installing this skill downloads the exact code at commit 1c8d6a68, not the current state of the repository. This prevents supply-chain attacks from unauthorized updates.
Security Audits
| Gen Agent Trust Hub | Pass |
| Socket | Pass |
| Snyk | Pass |
🌐 Community
Passed automated security scans.
Install command fetches the verified snapshot, not the live repository.