Precise Lite Models
November 4, 2025 ยท View on GitHub
Free models for usage with precise-lite
Related repos
- NEW - ovos-ww-plugin-precise-onnx - OVOS plugin for precise-lite (.onnx models)
- ovos-ww-plugin-precise-lite - OVOS plugin for precise-lite (.tflite models)
- precise-lite-trainer - train/convert models with this
- precise-lite-runner - standalone precise tflite model runner
- precise-community-data - Community data
Training
Model Info
| model | ww dataset | not-ww datasets | epochs | batch size | sensitivity | dopout | notes |
|---|---|---|---|---|---|---|---|
| hey_mycroft_mk1 | proprietary MycroftAI (user opt-in data) | user opt-in data (?) + hundreds of hours of youtube audio | 6000 | 5000 | 0.8 | 0.2 | model for the Mark I - 114,000 'hey mycroft' wake word samples, data collected via previous pocketsphinx wakeword and tagged via crowdsourcing in a WebUI |
| hey_mycroft | proprietary MycroftAI (user opt-in data) | 68,000 samples subset of hey_mycroft_mk1 dataset | 6000 | 5000 | 0.8 | 0.2 | model for the Mark II - balanced version of the original set, classified samples into high/low pitched (34,000 samples each). |
| christopher | proprietary MycroftAI | user opt-in data (?) + hundreds of hours of youtube audio (?) | unannounced but uploaded to MycroftAI github | ||||
| cough | proprietary MycroftAI | hundreds of hours of youtube audio (?) | unannounced but uploaded to MycroftAI github, probably the result of their SickWeather collaboration | ||||
| sheila | google-speech-commands | synthetic data + pdsounds + proprietary dataset + mycroft community dataset | 1000 | 3000 | 0.1 | ||
| marvin | google-speech-commands | synthetic data + pdsounds + proprietary dataset + mycroft community dataset | 1000 | 100 | 0.1 | ||
| android | proprietary dataset | synthetic data + pdsounds + google-speech-commands + mycroft community dataset | dataset known to contain mislabeled samples | ||||
| hey_robin | proprietary dataset | synthetic data + pdsounds + google-speech-commands + mycroft community dataset | 1000 | 3000 | 0.2 | ||
| hey_firefox | probably common voice (?) | pdsounds + google-speech-commands + mycroft community dataset + ? | no longer available, dataset was linked in castorini/howl repository | ||||
| hey_moxie | ? | pdsounds + google-speech-commands + mycroft community dataset + ? | |||||
| hey_scout | ? | pdsounds + google-speech-commands + mycroft community dataset + ? | |||||
| hey_chatterbox | mycroft community dataset | pdsounds + google-speech-commands + mycroft community dataset + ? | |||||
| computer | mycroft community dataset | pdsounds + google-speech-commands + mycroft community dataset + ? | 100 steps, .9938, 52 training words, 5 test | ||||
| hey_k9 | mycroft community dataset | pdsounds + google-speech-commands + mycroft community dataset + ? | |||||
| athena | mycroft community dataset | pdsounds + google-speech-commands + mycroft community dataset + ? | |||||
| ey_ordenador_0.2 | mycroft community dataset | pdsounds + google-speech-commands + mycroft community dataset + ? | 250 steps, 44 training words, 5 test | ||||
| ey_ordenador_0.3 | mycroft community dataset | pdsounds + google-speech-commands + mycroft community dataset + ? | 50 steps, 44 training words, 5 test |