"Identifying Different Musical Instrument Sounds Using Fourier Analysis in LabVIEW" Rather than "ML" du jour, I would say that a fast Fourier transform would get you sufficient data to determine if practicing or talking or silence.

Definitely valid callout. I was also looking for an excuse to play with audio ML, but you're totally right that just examining a Fourier series could very likely do a great job of determining whether a given type of instrument is being played.

Which an FFT on an ESP32-S3 with it's Xtensa LX7 vector extensions can be quite fast

https://www.reddit.com/r/arduino/comments/16asjx9/esp32s3_do...

The actual github : https://github.com/stolk/little_kiss_fft

The adversarial child would be playing recordings into the timer within minutes no?

I'm assuming the child is question is not so much adversarial so much as ADHD. They'll be actively looking to divert their attention to more stimulating activities than playing an instrument but not attempting to cheat the system. And I'm assuming a parent is present but not necessarily in full control of the child. Or it could work equally well for an adult in a similar situation.

For the non-adversary who wants something similar in nature, that might help them practice, why not? Not everyone is trying to cheat.

The ESP32S(1?) that's inside is fairly powerful as embedded mcus go but that said...

If you need grunt and you don't specifically want the aesthetic of an led matrix panel you'd probably be better off with an old phone or tablet based thing.

The TC001 afaik doesn't have any mics inside anyway.

Hate to burst your bubble, but as an amateur musician I fear this would backfire, or at least fail to result in any improvement in playing ability. Silence and time are absolutely critical to playing music. By analogy, measuring 'time spent drawing bow across strings' would be as useful to a violin student as 'time spent pressing foot on accelerator' would be to a driving student!

From my own experience learning to play the organ, I have improved least when I play relatively fluidly, practising with music well within my abilities. On the contrary, the most improvement has come when I've slowed down, allowed myself to count the timing, repeat sections, read the sheet music more carefully or even just take a break entirely. So although silence won't improve one's playing by itself, I think it's a natural by-product of an effective studying technique that, if at all possible, shouldn't be discouraged with such a timer.

There could be an allowance for time spent not playing music. Keep the timer going for a few minutes after the music has stopped. Require a minimum ratio of music playing time to overall time. So, after a lengthy silence, only playing music will advance the timer.

But, yeah, it may or may not be helpful.

It's a great point. For a contemplative learner deeply focused on what they're doing, this is definitely a bad idea. I'm aiming at sort of the opposite kind of learner.

i'm building a low volume product. if you add the cost unfortunately you will see it's impossible to do something of good quality and low volume with much lower price.

you don't only add the sum of the electronic parts, but materials, r&d, payment for the one who does the assembly, marketing, taxes & accounting, shipping package, refunds and defects, payment for someone to produce it or the usage of your existing equipment, etc. if you add all of these, you will see that it's not cheap.