StuartIanNaylor

controls [ 84 -10 30 20 0.05 ] 0.2 too high Still seems to cut a tiny last portion of the ending samples controls [ 84 -15 30 20 0.05...

PS if you want to check with a KW model here is the one I have just been playing with ``` import tensorflow as tf import sounddevice as sd import...

PS only a tiny thing and purely for info as it works but left over time the gain creeps up so that you clip the 1st command. Haven't really worked...

Yeah you have a protobuf version requirement

If you wipe the firmware and have it as just a snapcast client then prob yeah if you can work out the i/o and stuff. If your thinking it could...

I had watchdog working the other day so its not 2.6. Did find it a bit funny peculiar though and async stuff like that twists my melon. Watchdog was working...

Last time I looked was quite a bit before Buster came out so apols

Yeah prob can by giving you a look at https://github.com/usefulsensors/openai-whisper as [nyadla-sys](https://github.com/usefulsensors/openai-whisper/commits?author=nyadla-sys) has created various examples for exports to onnx and then tflite, so plenty of templates for the conversion...

@fakufaku https://github.com/wenet-e2e/wesignal as might be of interest to what is submitted

https://github.com/SaneBow/PiDTLN