wavecap-whisper
Tune WaveCap Whisper model settings. Use when the user wants to change model size, adjust decoding parameters, configure prompts, or optimize transcription accuracy vs speed.
npx skills add majiayu000/claude-skill-registry --skill wavecap-whisper-tobiaswooldridge-wavecap --agent claude-code
Same command for any agent — swap --agent for codex, cursor, copilot.
Weekly change comes from our own snapshots, not the repository page — it measures attention, not adoption.
# WaveCap Whisper Model Tuning Skill Use this skill to tune the Whisper speech-to-text model for optimal transcription quality. ## Configuration Location Whisper settings are in the `whisper:` section: - **User config:** `/Users/thw/Projects/WaveCap/state/config.yaml` - **Default config:** `/Users/thw/Projects/WaveCap/backend/default-config.yaml` ## Model Selection ### Primary Model ```yaml whisper: model: large-v3-turbo # Model checkpoint ``` | Model | Size | Speed | Accuracy | Use Case | |-------|------|-------|----------|----------| | tiny | 39M | Fastest | Low | Testing only | | base | 74M | Very fast | Fair | CPU fallback | | small | 244M | Fast | Good | Limited GPU | | medium | 769M | Moderate | Very good | Balanced | | large-v2 | 1.5B | Slow | Excellent | High accuracy | | large-v3 | 1.5B | Slow | Best | Maximum accuracy | | **large-v3-turbo** | 809M | Fast | Excellent | **Recommended** | ### Backend Selection ```yaml whisper: backend: auto # auto, mlx, faster-whisper ``` | Backend | Platform | Performance | |---------|----------|-------------| | auto | Any | Detects best option | | mlx | Apple Silicon | 10-50x faster than CPU | | faster-whisper | CUDA/CPU | Best for NVIDIA
- Configuration Location
- Model Selection
- Primary Model
- Backend Selection
- CPU Fallback Model
- Decoding Parameters
- Beam Size (search width)
- Temperature (randomness)
- Condition on Previous Text
- Language Configuration
- Initial Prompts (domain vocabulary)
- Global Prompt
- Named Prompts (reusable)
- Assign to Stream
grep -A30 "whisper:" /Users/thw/Projects/WaveCap/state/config.yaml | head -35
curl -s http://localhost:8000/api/health | jq
launchctl stop com.wavecap.server && sleep 2 && launchctl start com.wavecap.server
curl -s http://localhost:8000/api/transcriptions/export | \
jq 'group_by(.confidence | . * 10 | floor / 10) |
jq '[.[] | select(.confidence < 0.7)] | sort_by(.confidence) | .[:5] | .[] | {confidence, text}'What does the wavecap-whisper skill do?
Tune WaveCap Whisper model settings. Use when the user wants to change model size, adjust decoding parameters, configure prompts, or optimize transcription accuracy vs speed.
How do I install it?
Run `npx skills add majiayu000/claude-skill-registry --skill wavecap-whisper-tobiaswooldridge-wavecap --agent claude-code` — it drops the skill into your project so the agent can pick it up. Swap the --agent value for codex, cursor or copilot if you use one of those.
Where does this skill come from?
From majiayu000/claude-skill-registry, a repository with 534 stars. We read it straight from the repository tree rather than a submitted listing, so what you see here is what is actually published.
Is a popular skill a good skill?
Not necessarily. Stars measure attention, not adoption — a repository can trend for a week and be abandoned. That is why we show the weekly change from our own snapshots next to the total, instead of a single flattering number.