Skip to content

Open-source AI podcast clipper. Generate vertical clips with face tracking and burned-in captions. CLI, MCP server, web app.

1.0 KB
Raw Blame
# Transcription
openai-whisper>=20250625

# Speaker diarization (optional: only needed if you enable local speaker detection)
# Most users don't need this: Claude handles speaker identification via MCP,
# and the "Paste Transcript" flow supports Speaker (MM:SS) labels natively.
# To enable: pip install pyannote.audio speechbrain && set HF_TOKEN in .env
# pyannote.audio>=3.1.0
# speechbrain>=1.0.0

# Face detection (optional, for smart cropping)
opencv-python-headless>=4.8.1.78
numpy>=2.5.2

# Audio-event detection (laughter/reaction channel via YAMNet ONNX, no torch/TF)
onnxruntime>=1.28.0

# --engine omnilingual: Meta's Omnilingual ASR CTC model via sherpa-onnx, for
# languages whisper.cpp handles poorly (e.g. Georgian). No torch/TF.
sherpa-onnx>=1.13.8

# Thumbnails
Pillow>=12.3.0

# CLI interactive prompts
questionary>=2.1.1

# Utilities
python-dotenv>=1.2.2
yt-dlp>=2026.7.4

# YouTube analytics (optional: only needed for live OAuth sync; CSV import works without these)
google-api-python-client>=2.198.0
google-auth-oauthlib>=1.4.0