How can train using my own dataset

RetroCirce / Zero_Shot_Audio_Source_Separation

The official code repo for "Zero-shot Audio Source Separation through Query-based Learning from Weakly-labeled Data", in AAAI 2022

https://arxiv.org/abs/2112.07891

MIT License

186 stars 31 forks source link

How can train using my own dataset #22

Open Hope1337 opened 10 months ago

Hope1337 commented 10 months ago

Same as title : How can train using my own dataset ?

RetroCirce commented 9 months ago

Hi, sorry for the late reply.

You need to revise or refer to the data_processor.py file to change the dataset loader and dataset classes. I use the "LGSPDataset" module to process the audioset data, so if you want to use our dataset like a similar way. You can take a look on that.

If you already get the separation training data (i.e., both mixture and separation/source tracks), things might become more easier because you don't need mix the data but directly trained with targets.

Hope1337 commented 9 months ago

I'll try, thanks <3