mwild1
commented
2 months ago This is interesting news, thanks for sharing!
I will note though, that it sounds like their focus is on voice assistants (understandable, since that is a popular topic right now). However my interest in this issue (and why I am tracking it) goes beyond voice. For example, I want to use esphome to detect my smoke alarm sounding (I believe some of the commercial devices can do this, I think Google Home for example). In theory this shouldn't be too hard to do, but plugging it all together appropriately in esphome isn't as easy as it could be.
Describe the problem you have/What new integration you would like A VU meter and VAD (voice activity detection) available to the i2s_audio microphone and eventually other means of signal processing for audio. Available within in the Arduino framework for ESP32.
Please describe your use case for this integration and alternatives you've tried: The current state of the 'voice_assistant' is sub-par. VAD and silence detection rely on the ESP-ADF which isn't free software and are implemented inside of the 'voice_assistant' component. I'd like a seperate component available to do audio signal processing. A VU meter / silence detection / voice activity detection and we could also implement clap detection, whistle detection and stuff like that. Maybe later on noise suppression and wake-word detection. This could all seperated into one component.
Additional context
Is there an empty component / similar component / boilerplate code available to get me started with a component like that? I've read the documentation on components but there is a steep learning curve until you get the boilerplate-code usable.
I'm not aware of any good signal processing libraries. ArduinoSound has some simple examples available, and there is arduino-audio-tools.
And can we use the ESP-DSP? It's available both for the ESP-IDF and as part of the ESP32 Arduino framework. It contains things like FFT. Or is this too custom because it only works for ESP32?