transcribe_audio
Installation
SKILL.md
ONNX-ASR Transcription Skill
This skill provides practical, local speech-to-text capabilities using onnx-asr.
🎯 Menu of Models & Use Cases
Choose the right model based on the use case. By default, use nemo-parakeet-tdt-0.6b-v2 for general English.
| Model | Best For | Characteristics |
|---|---|---|
nemo-parakeet-tdt-0.6b-v2 |
General purpose English | High accuracy, fast, requires ~2.5GB RAM |
nemo-canary-1b-v1 |
Multi-lingual & Translation | Supports English, Spanish, German, French. Slower, higher RAM. |
whisper-base.en |
Low resource English | Very fast, lower accuracy, ~500MB RAM |
whisper-small |
Low resource Multi-lingual | Balanced speed/accuracy for multiple languages, ~1GB RAM |