Failed to loading frames/audio

[krishnareedy]

Sir, first i created .csv files, in the csv files it is showing what inputs are there and it's paths also. but during training it is showing failed to load frames/audio.

[krishnareedy]

[avis-ma]

Hello, I have the same problem. Have you sovled it already? And how to solve this problem? Thank you very much.

[avis-ma]

@krishnareedy

[AIaiAIaiAIaiAI]

@krishnareedy

[JusperLee]

I know what I ca n’t read the audio file, because when running base.py your path is "./data / ....", you can't find the data, or you should replace "./data" to "../Data", or use an absolute path.

[JusperLee]

Sorry, my statement above is not entirely correct. This repository uses the torchaudio method, and the value returned by the load function is [channel, lenth]. Therefore, we need to modify the load function of the base script and change shape [1] == 2 to shape [0] == 2. The following are similar.

[parth1497]

@JusperLee Thank you so much!! :D

[vipulSharma18]

@parth1497 I'm working on at least duplicating the results of this project since you are also working on it, maybe we can solve each other's doubts. There seem to be common issues like these that everyone doing this project has faced.

[TaoStarlit]

in my preprocess, the audio has been converted to mono, so the shape is (1,xxxx), which is not affected by the shape[0] or shape[1]. However, there are still so many: could not broadcast shape 0 to shape (xxx)

[2023zy]

Sorry, my statement above is not entirely correct. This repository uses the torchaudio method, and the value returned by the load function is [channel, lenth]. Therefore, we need to modify the load function of the base script and change shape [1] == 2 to shape [0] == 2. The following are similar.

@JusperLee Hello, I modified the load function of the base script as the picture showed , although the failure gets fewer, there is still some failure during training. I wonder if you know the reason for it . Thank you so much!