jayg996 / btc-ismir19 Goto Github PK

Hi, thanks for making this! I tested it, it's amazing. I wonder if it's possible to extract vectors for the wave files and if so is there example code for it?

May I know what version of python did you used in this case since I'm having an error
Traceback (most recent call last):
File "test.py", line 22, in
config = HParams.load("run_config.yaml")
File "C:\Users\LENOVO\AppData\Local\Programs\Python\Python37\BTC-ISMIR19\utils\hparams.py", line 29, in load
return cls(**yaml.load(f))
TypeError: load() missing 1 required positional argument: 'Loader'

Hello, really interesting work! I am curious what GPUs and how many of them are required for training BTC? And how much time does it converge? Thank you.

Can you please provide the details how can I arrange the dataset to execute your code?

In case it helps anyone else, I got this error while running test.py on a machine w/o CUDA support.

Attempting to deserialize object on a CUDA device but torch.cuda.is_available() is False. If you are running on a CPU-only machine, please use torch.load with map_location=torch.device('cpu') to map your storages to the CPU.

I fixed the problem by replacing the following line 29 in test.py:

checkpoint = torch.load(model_file)

with

torch.load(model_file, map_location=lambda storage, loc: storage)

Hi, thanks for your work! It seems that this code is for audio2chord experiments. How to run it on the BPS-FH dataset(i.e. symbolic music chord recognition)?

This is not a bug report.
I'm trying to relearn(train.py) BTC-ISMIR2019, but I'm not sure about the directory structure under /data/music/chord_recognition/
Would you tell me where to put the mp3 and lab files?
P.S.
I already have the sound source data (.wav or .mp3) and annotation data (.lab) of isophonic, USpop, and robbie williams.