mhu-coder/ConvTasNet_Libri1Mix_enhsingle

Description:

This model was trained by Mathieu Hu using the librimix/ConvTasNet recipe in Asteroid. It was trained on the enh_single task of the Libri1Mix dataset.

 

Training config:

• data:
  • n_src: 1
  • sample_rate: 16000
  • segment: 3
  • task: enh_single
  • train_dir: data/wav16k/min/train-100
  • valid_dir: data/wav16k/min/dev
• filterbank:
  • kernel_size: 16
  • n_filters: 512
  • stride: 8
• main_args:
  • exp_dir: exp/train_convtasnet_f34664b9
  • help: None
• masknet:
  • bn_chan: 128
  • hid_chan: 512
  • mask_act: relu
  • n_blocks: 8
  • n_repeats: 3
  • n_src: 1
  • skip_chan: 128
• optim:
  • lr: 0.001
  • optimizer: adam
  • weight_decay: 0.0
• positional arguments:
• training:
  • batch_size: 2
  • early_stop: True
  • epochs: 200
  • half_lr: True
  • num_workers: 4

 

Results:

• si_sdr: 13.938355526049932
• si_sdr_imp: 10.488574220190232
• sdr: 14.567380104207393
• sdr_imp: 11.064717304994337
• sir: inf
• sir_imp: nan
• sar: 14.567380104207393
• sar_imp: 11.064717304994337
• stoi: 0.9201010933251715
• stoi_imp: 0.1241812697846321

 

Versions:

• torch_version: 1.5.0
• pytorch_lightning_version: 0.7.6
• asteroid_version: 0.2.1

 

License notice:

This work "ConvTasNet_Libri1Mix_enhsingle" is a derivative of LibriSpeech ASR corpus by Vassil Panayotov, used under CC BY 4.0; of The WSJ0 Hipster Ambient Mixtures dataset by Whisper.ai, used under CC BY-NC 4.0 (Research only). "ConvTasNet_Libri1Mix_enhsingle" is licensed under Attribution-ShareAlike 3.0 Unported by Mathieu Hu.