https://github.com/Daenara/rhasspy_save_wakeword

Saves wakewords from mqtt audio stream once detected
https://github.com/Daenara/rhasspy_save_wakeword

Last synced: 2 months ago
JSON representation

Saves wakewords from mqtt audio stream once detected

• Host: GitHub
• URL: https://github.com/Daenara/rhasspy_save_wakeword
• Owner: Daenara
• License: mit
• Created: 2021-05-24T23:57:19.000Z (over 3 years ago)
• Default Branch: master
• Last Pushed: 2021-05-25T00:25:40.000Z (over 3 years ago)
• Last Synced: 2024-08-03T16:09:39.642Z (6 months ago)
• Language: Python
• Homepage:
• Size: 4.88 KB
• Stars: 2
• Watchers: 2
• Forks: 0
• Open Issues: 0
• Metadata Files:
  • Readme: README.md
  • License: LICENSE

Awesome Lists containing this project

• awesome-rhasspy - rhasspy_save_wakeword - Tool to collect wake word samples from the Hermes MQTT audio stream for custom model training. (Miscellaneous)

README

        
# rhasspy_save_wakeword

This is a small script that listens to the hermes/audioServer/+/audioFrame and hermes/hotword/+/detected mqtt topics and saves the audio it buffers from audioFrame after the wakeword was detected. 

The intended use is to collect wakeowrd samples (and wrongly detected wakewords) for custom model training.

Requirements are paho-mqtt and pydub. This was only tested on python 3.7