iver56 / clap-detection Goto Github PK

With this feature, one can have different clap rhythms that can trigger different actions. For example, one specific clap rhythm will turn on the lights while a different clap rhythm will turn on the TV.

As a musician, one might be more used to 'length of notes', instead of 'length of pauses'.

So in the example notes given

Instead of defining the pattern to be [2, 1, 1, 2] as the relative pauses between 'notes',
you could define it in musical beats as [1./4, 1./8, 1./8, 1./4, 1./4].

Now, a 'converter' could find the smallest of these fractions, and do multiplications to convert it to the proper pattern format.

Just looking for transients is not good enough because it results in a lot of false positives. A better approach would be machine learning

• Discard background noise. Sample the noise while the spectrum is not in flux.
• Train a neural network to recognize clap sounds but reject other types of sounds with a transient.

Connect with If This Then That

Hi

I have a cheap microfone usb-stick but the input level is quite low.
In alsamixer I can go up to 23.81dB but its still to silent to recognize
my claps. Is it possible to somehow preprocess the input stream somehow
that it gets amplified? Is it correct to set 0dbfs = 23 in clap.csd?

I just encountered an input error with my Aureon Dual sound card.
unable to set number of channels...it said.

So I just wonder which sound card you use. And do you have any recommendation on sound card that works fine on raspberry pi?