Closed darkzbaron closed 3 years ago
Unfortunately, this is not enough information to help fixing your issue. E.g., are you able to record something by yourself with pyaudio?
Hi thanks so much for the rep. I was using an old bin file for DBNBeatTracker which caused an issue while using the latest madmom package.
It looks like you remove the ability to select the recording pyaudio device! It's a shame now I have to modify the source.
Also, any idea what would be the most appropriate algorith to animate a light to music? I am using the DBNBeatTracker but there are "not enough" beat events to my tast, I tried to increase the gain but not sure that's the right approach.
Thanks so much for the great work anyway
Regarding the ability to select the audio source: this should still be there (see the --list
option of all online capable scripts to list all devices selectable with --device
).
You might want to set --min_bpm
to a higher value, this should force the DBN to chose a higher tempo, most likely double of the correct tempo, but please make sure to increase --max_bpm
accordingly.
If this does not work, you could try using OnsetDetectorLL
, but this gives all onsets of an audio, which might not be very periodic and thus less suited to control lights.
HI there, thanks so much for your help! I am making progress using DBNBeatTracker,. I am looking to add another dimension to my script, namely the intensity of the beat, can I get that from your source or does the DBN only returns confidence level/probability?
The DBN returns only the positions of the beats, but you can of course re-evaluate the output of the RNN. But maybe a simple thresholding of the RNN activation is enough for your apllication.
Expected behaviour
I would have expected Madmom to use the default input device
Actual behaviour
No information on actual input device used and nothing happen
Steps needed to reproduce the behaviour
Information about installed software
Python 37
` madmom 0.16.1 numpy 1.20.2 PyAudio 0.2.11