yeyupiaoling/AudioClassification-Pytorch

can flac format be used as dataset

huangp0600 opened this issue · 3 comments

Great thanks for providing such an awesome project !
I have a batch of audio files with flac format, can they used as dataset to run this project

@huangp0600 Of course. Support wav, flac, mp3 .....

@yeyupiaoling great thanks for quick reply.
I have another question about model training.
case1
I have 100 audio files, each with a duration of 20s
case2
I split each audio file to two parts. then I have 200 files totally, each with a duration of 10s.

which is better for training model, or which can achieve better accuracy?

@huangp0600 case2
During training, the audio will be cropped automatically, but if you have a more reasonable cropping, it will be more beneficial to train the model.