Micro Wake Word Trainer Apple Silicon – microWakeWord-Trainer-AppleSilicon v1.0

Written by

Tater Totterson

in

Micro Wake Word Trainer Apple Silicon – microWakeWord-Trainer-AppleSilicon v1.0

Hey AI Enthusiasts! 👋

Great news for all you Home Assistant voice control fans! microWakeWord-Trainer-AppleSilicon v1.0 is here – a fully automated wake-word trainer optimized for Apple Silicon (M1-M4) Macs! 🍏

Here’s the scoop:

  • Simplified Training: One bash command (`./train_microwakeword_macos.sh “wake word”`) does it all – TTS sample generation with Piper, training, and exporting a quantized model for ESPHome.
  • GPU Accelerated: Uses TensorFlow-Metal to harness the power of your Apple Silicon chip for faster training!
  • Offline Functionality: Local caching ensures smooth operation even without a HuggingFace connection.
  • Natural Sounding Wake Words: Powered by Piper TTS for realistic voice samples.
  • ESPHome Integration: Creates `.tflite` models and `.json` metadata perfect for your ESPHome projects.

Check it out & start training: https://github.com/TaterTotterson/microWakeWord-Trainer-AppleSilicon/releases/tag/1.0

Happy building! 🥔☕️

https://github.com/TaterTotterson/microWakeWord-Trainer-AppleSilicon/releases/tag/1.0

More posts