search

audiojs / contributing

Discussion and guidelines for contributing
2 stars 0 forks source link

Passing tests on travis #13

Closed jamen closed 7 years ago

jamen commented 8 years ago

Right now most of the tests don't pass on Travis. Here are the statuses: https://travis-ci.org/audiojs

jamen commented 8 years ago

looks like i might have done the configs wrong on a few? some say ruby. I was using generate-travis for them all. That is just audio-shader, forgot the config.

dy commented 8 years ago

Interesting what is due way to assert results. Too precise comparison is no good, due to different delays, glitches etc. But too loose neither, in case if channels swapped or noise instead of a sine appeared.

Maybe using loose spectrum estimation would work, with fourier-transform. Or even better - detect N spectral peaks and use almost-equal for magnitudes/frequencies comparison.

Should think about audio-fft component, to convert audio data to frequencies data, maybe. Also some sort of peaks detector.

It is a bit different subject though.