wavecap-llm
Configure WaveCap LLM-based transcription correction. Use when the user wants to enable/disable LLM correction, change models, tune prompts, or optimize correction quality on Apple Silicon.
npx skills add majiayu000/claude-skill-registry --skill wavecap-llm-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 LLM Correction Tuning Skill Use this skill to configure the optional LLM-based post-processing that corrects Whisper transcription errors using local models on Apple Silicon. ## Requirements - Apple Silicon Mac (M1/M2/M3/M4) - mlx-lm package installed: `pip install mlx-lm` - Sufficient RAM for model (1B=2GB, 3B=6GB, 8B=16GB) ## Configuration Location LLM settings are in the `llm:` section: - **User config:** `/Users/thw/Projects/WaveCap/state/config.yaml` ## Basic Configuration ### Enable/Disable LLM Correction ```yaml llm: enabled: true # false to disable ``` ### Model Selection ```yaml llm: model: llama-3.2-3b # Model name or HuggingFace path ``` | Model | Size | RAM | Speed | Quality | Use Case | |-------|------|-----|-------|---------|----------| | llama-3.2-1b | 1B | ~2GB | Fastest | Good | Low RAM, quick fixes | | qwen-2.5-1.5b | 1.5B | ~3GB | Very fast | Good | Balanced small model | | **llama-3.2-3b** | 3B | ~6GB | Fast | Very good | **Recommended** | | qwen-2.5-3b | 3B | ~6GB | Fast | Very good | Alternative to Llama | | llama-3.1-8b | 8B | ~16GB | Moderate | Excellent | High quality | | llama-3.2-8b | 8B | ~16GB | Moderate | Excellent | Latest 8B | ### Generatio
- Requirements
- Configuration Location
- Basic Configuration
- Enable/Disable LLM Correction
- Model Selection
- Generation Parameters
- Domain Terms
- View Current Settings
- Check LLM Status
- Full Configuration Example
- Tuning Scenarios
- Maximum Quality (8GB+ RAM)
- Balanced (6GB RAM)
- Low Memory (4GB RAM)
grep -A20 "^llm:" /Users/thw/Projects/WaveCap/state/config.yaml
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 '[.[] | select(.llmCorrectedText != null and .llmCorrectedText != .text)] |
jq '{
ls -la ~/.cache/huggingface/hub/ | grep mlx
tail -50 /Users/thw/Projects/WaveCap/state/logs/backend.log | grep -i llmWhat does the wavecap-llm skill do?
Configure WaveCap LLM-based transcription correction. Use when the user wants to enable/disable LLM correction, change models, tune prompts, or optimize correction quality on Apple Silicon.
How do I install it?
Run `npx skills add majiayu000/claude-skill-registry --skill wavecap-llm-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.