xiao-ming / chordrecognitionmiditrainedextractor Goto Github PK

Hello,
I've been trying to use the code you provided on this repository with a dataset of around 500 songs. When retraining the network I receive an out of memory exception from CuPy. From some calculations I made based on the amount of songs and the associated memory usage, it seems I'd require around 500Gb of GPU memory to train the network with my dataset. Are you aware of this issue? Is there any workaround for it, or a way to reduce memory usage when training?

Hello!
I've tried the code for training the decoder by my own audio datasets,
and successfully print out the estimation label.
I would like to know if the "evaluation.py" file is for calculating its accuracy, and how the 3 functions work(the parameters) for further, hope to hear from you! Thank you!

Hi, is there any way to convert .lab file to midi file?

This source loss blstm.model, when run Estimate_Chords_From_Audio.py , it will give a error info.