Good Morning (UTC +10),
I am playing with a couple of PE’s I brought (end result is to replace all the very old google units around the house), I am setting up “Hey Jarvis” for a mate but to keep our houses different I want to go Cortana as my wake word (that is next on the hit list). What i wanted to do is change the pre announce sound to “Master Chief”, i have the recording as a high quality flac (26kb) and 770ms long (i know its long but hey).
To be able to fiddle i downloaded the code for 2025.3.3 and loaded it manually into ESP Home and changed the wake_word flac to that off my home assistant www directory.
First triggering works perfectly (after every reboot), not an issue (had to add a 820ms (770ms + 50ms buffer) delay between playing the sound and accepting incoming input as it heard itself say “Master Chief”), but subsequent triggers cut the audio about 20ms short and added a small distortion to the sound, still works but annoying. Trying to figure out why though. If any experts could take a look and see where i went wrong or make suggestions?
i used this code - home-assistant-voice-pe/home-assistant-voice.yaml at 95c83c74f777ddcd573ff21d4137d25f9b621b02 · esphome/home-assistant-voice-pe · GitHub
and made these modifications
- id: wake_word_triggered_sound
file: http://192.168.69.8:8123/local/wake_word_triggered.flac
under micro_wake_word
else:
- if:
condition:
switch.is_on: wake_sound
then:
- script.execute:
id: play_sound
priority: true
sound_file: !lambda return id(wake_word_triggered_sound);
- delay: 820ms
- delay: 820ms
- voice_assistant.start:
wake_word: !lambda return wake_word;
- The delays run in parallel, there is only a 820ms wait from activation to when it accepts audio, just long enough to play the sound.