bold imp
Super confused here. My ESP32 based satellite recognizes wake word, listens, but doesnt detect that I am done speaking. (I guess?) It just sits there.
Note: I do not have a speaker configured. (I dont want audio feedback, just visual [light]) I wouldn't think it's required as thee browser-based assist works the same way.
I could use some help.
Logs from ESP attached
relevant ESP config
switch:
- platform: template
id: mute
name: mute
optimistic: true
on_turn_on:
- micro_wake_word.stop:
- voice_assistant.stop:
- light.turn_on:
id: led_ww
red: 100%
green: 0%
blue: 0%
brightness: 60%
effect: fast pulse
- delay: 2s
- light.turn_off:
id: led_ww
- light.turn_on:
id: led_ww
red: 100%
green: 0%
blue: 0%
brightness: 30%
on_turn_off:
- micro_wake_word.start:
- light.turn_on:
id: led_ww
red: 0%
green: 100%
blue: 0%
brightness: 60%
effect: fast pulse
- delay: 2s
- light.turn_off:
id: led_ww
light:
- platform: esp32_rmt_led_strip
id: led_ww
rgb_order: GRB
pin: GPIO21
num_leds: 1
chipset: ws2812
name: "on board light"
effects:
- pulse:
- pulse:
name: "Fast Pulse"
transition_length: 0.5s
update_interval: 0.5s
min_brightness: 0%
max_brightness: 100%
# Audio and Voice Assistant Config
i2s_audio:
- id: i2s_in
i2s_lrclk_pin: GPIO2 #WS
i2s_bclk_pin: GPIO3 #SCK
microphone:
- platform: i2s_audio
id: va_mic
adc_type: external
i2s_din_pin: GPIO4 #SD pin on the INMP441
channel: left
pdm: false
i2s_audio_id: i2s_in
bits_per_sample: 32 bit
on_wake_word_detected:
# then:
- logger.log: "🎤 Wake word triggered!"
- voice_assistant.start:
wake_word: !lambda return wake_word;
- light.turn_on:
id: led_ww
red: 30%
green: 30%
blue: 70%
brightness: 60%
effect: fast pulse
models:
- model: github://esphome/micro-wake-word-models/models/v2/hey_jarvis.json@main
vad:
model: github://esphome/micro-wake-word-models/models/v2/vad.json@main
voice_assistant:
id: va
microphone: va_mic
noise_suppression_level: 4.0
volume_multiplier: 4.0
on_stt_end:
then:
- light.turn_off: led_ww
on_error:
- micro_wake_word.start:
on_end:
then:
- light.turn_off: led_ww
- wait_until:
not:
voice_assistant.is_running:
- micro_wake_word.start:
As you can see, it successfully hear's the wake word, streams the mic, but never determines that it's time to stop.
bold imp
update:
I have now added a speaker and media_player components to the ESPHome config, with no change. wake word is detected, HA kinda sees that the pipeline was opened, but no events are recorded.