@hasna/transcriber

Transcribe audio and video from files and URLs (YouTube, Vimeo, Wistia, etc.) using ElevenLabs, OpenAI Whisper, or DeepGram. Full CLI + MCP server.

Install

bun install -g @hasna/transcriber
# or
npm install -g @hasna/transcriber

Data is stored at ~/.hasna/transcriber/transcriber.db. Override with HASNA_TRANSCRIBER_DIR or TRANSCRIBER_DIR.

CLI Usage

# Transcribe a YouTube video
transcriber transcribe https://youtube.com/watch?v=...

# Transcribe a local file with OpenAI Whisper
transcriber transcribe /path/to/audio.mp3 --provider openai

# Transcribe with diarization (speaker detection) — ElevenLabs only
transcriber transcribe https://youtube.com/watch?v=... --diarize

# List transcripts
transcriber list

# Get a transcript
transcriber get <id>

# Search transcripts
transcriber search "machine learning"

# Export as SRT subtitles
transcriber export <id> --format srt

# Summarize with AI
transcriber summarize <id>

# Translate to another language
transcriber translate <id> --to fr

# Proofread
transcriber proofread run <id>

# Annotate at timestamp
transcriber annotate add <transcript-id> --at 123 --note "Key moment"

# Check providers
transcriber providers

MCP Server

transcriber-mcp

Add to your Claude/MCP config:

{
  "mcpServers": {
    "transcriber": {
      "command": "transcriber-mcp"
    }
  }
}

Providers

Requirements

• yt-dlp — for downloading YouTube/Vimeo/etc.
• ffmpeg — for audio processing and clip export

Features

• Transcribe files and URLs (YouTube, Vimeo, Wistia, and any yt-dlp source)
• Speaker diarization (identify multiple speakers)
• Word-level timestamps
• Auto-chunk long audio (>10 min)
• AI summarization (OpenAI / Anthropic)
• Translation to any language
• Proofread issues (spelling, grammar, punctuation, clarity)
• Export: TXT, SRT, VTT, ASS, Markdown, JSON
• Timestamped annotations/bookmarks
• YouTube comment import
• Podcast RSS feed auto-transcription
• Notion export
• Webhook notifications
• Tag-based organization

License

Apache-2.0