AgentToolHub
Skill

faster-whisper

faster-whisper — a Claude Code skill from theplasmak/faster-whisper

Features

  • ✓**~4-6x faster** than OpenAI's original Whisper (same model weights, CTranslate2 backend)
  • ✓**~20x realtime** with GPU — transcribe 10 min of audio in ~30 sec
  • ✓**Distilled models** available (~6x faster again with <1% WER loss)
  • ✓**Speaker diarization** (`--diarize`) — labels who said what via pyannote.audio
  • ✓**GPU acceleration** (NVIDIA CUDA) with quantization for CPU efficiency
  • ✓**YouTube/URL input** — auto-downloads via yt-dlp (YouTube, direct links, etc.)
  • ✓**Subtitle output** — SRT, VTT, ASS, LRC, TTML, HTML formats
  • ✓**Word-level timestamps** with automatic wav2vec2 alignment (~10ms precision)
  • ✓**Voice activity detection (VAD)** — removes silence automatically (on by default)
  • ✓**Audio preprocessing** — normalize volume (`--normalize`) and denoise (`--denoise`)
  • ✓**Filler word removal** (`--clean-filler`) — strip um/uh/er/ah/hmm and discourse markers
  • ✓**Transcript search** (`--search TERM`) — find all timestamps where a word/phrase appears
  • ✓**Chapter detection** (`--detect-chapters`) — auto-detect sections from silence gaps
  • ✓**CSV/TSV output** — spreadsheet-ready transcripts with proper quoting
  • ✓**Paragraph detection** (`--detect-paragraphs`) — insert natural paragraph breaks in text output
  • ✓**Podcast RSS transcription** (`--rss URL`) — fetch and transcribe episodes from a feed
  • ✓**Stereo channel selection** (`--channel left|right`) — transcribe a specific stereo track
  • ✓**Speaker audio export** (`--export-speakers DIR`) — save each speaker's turns as separate WAV files
  • ✓**Batch processing** with automatic ETA — glob patterns, directories, skip-existing
  • ✓**Translation to English** (`--translate`) — any language → English
  • ✓**Multilingual support** (99+ languages with auto-detection)

Installation Guide

Option 1: Install from ClawHub

Via CLI (no installation required):

bash
# Using npx (npm)
npx clawdhub@latest install faster-whisper

# Using pnpm
pnpm dlx clawdhub@latest install faster-whisper

# Using bun
bunx clawdhub@latest install faster-whisper

This downloads and installs the skill into your default skills directory (~/clawd/your-agent/workspace/skills/ or similar).

Via Web UI:

  • Go to https://clawdhub.com/ThePlasmak/faster-whisper

  • Download the zip

  • Unzip the zip and move the contents into a faster-whisper folder in your skills directory.

    • Option 2: Download from GitHub Releases

  • Go to Releases

  • Download the latest faster-whisper-X.X.X.zip

  • Extract it to your agent's skills folder:

    • - Default location: ~/clawd/your-agent/workspace/skills/faster-whisper
    - Or wherever your agent's workspace is configured

    bash
    # Example
cd ~/clawd/your-agent/workspace/skills/
unzip ~/Downloads/faster-whisper-1.0.1.zip -d faster-whisper

    If you're lazy: You can also ask your agent to install it by pasting this repo's link (https://github.com/ThePlasmak/faster-whisper) directly in the chat.

    Note: The release zip excludes certain explanatory files (CHANGELOG, LICENSE, README) and only contains the skill itself — this keeps things lightweight.

    Reviews