[Suggestion] "FFT bands" as midi output?

[bestdani]

Hello,

I’m currently experimenting with the included MIDI for lights script and while the VU meter output is cool already I’m wondering if it would be easily possible to output “FFT audio spectrum bars” directly from mixxx in one of the next versions via MIDI of the currently playing track or maybe even the master.

I’d expect that at some stage something like this might be calculated anyway somewhere in the audio chain and therefore could be just made available for MIDI readout without much hassle. Besides using this for more control in lighting software directly from mixxx without rerouting the mixxx signal through lighting software where this would be calculated with some delay, I think this could maybe be interesting for other parts, at least some external mixers have such visualized so it could be maybe implemented in (future) MIDI hardware, too. Last but not least this could also serve as a little gimmick for some skins or maybe even as some real visual guide when mixing with EQs.

[kazakore]

I assume the analysed waveforms must have some level of fft frequency bands encoded in them as different frequencies are coloured and the waveform changes with the EQs. Maybe some way to export current position as read from this analysis data would work without being too processor hungry??

[Be0]

@kazakore

I assume the analysed waveforms must have some level of fft frequency bands encoded in them as different frequencies are coloured and the waveform changes with the EQs. Maybe some way to export current position as read from this analysis data would work without being too processor hungry??

Yes I think that would be feasible. It would be a bit of work to implement that of course.

[JosepMa]

AFAIR, the analysis only stores three bands for each step (which is each 1/100th of a second), bass, mid and treble, so no FFT as we’re used to.

[kazakore]

@JosepMa

AFAIR, the analysis only stores three bands for each step (which is each 1/100th of a second), bass, mid and treble, so no FFT as we’re used to.

That may still be useful with controlling lights and would be less resource intensive than sending the audio and doing a more complete spectral analysis at the other end…

[markus_b]

@kazakore

@JosepMa

AFAIR, the analysis only stores three bands for each step (which is each 1/100th of a second), bass, mid and treble, so no FFT as we’re used to.

That may still be useful with controlling lights and would be less resource intensive than sending the audio and doing a more complete spectral analysis at the other end…

Yes, I agree. That we don’t need an fft and its associated CPU load is also an advantage. Also this is sufficient for some nice additional lighting effects.