tmirzaev-dotcom/ConvTasNet_Libri3Mix_sepnoisy

Description:

This model was trained by TakhirMirzaev using the librimix/ConvTasNet recipe in Asteroid. It was trained on the sep_noisy task of the Libri3Mix dataset.

 

Training config:

• data:
  • n_src: 3
  • sample_rate: 8000
  • segment: 3
  • task: sep_noisy
  • train_dir: data/wav8k/min/train-360
  • valid_dir: data/wav8k/min/dev
• filterbank:
  • kernel_size: 16
  • n_filters: 512
  • stride: 8
• main_args:
  • exp_dir: exp/train_convtasnet_my_tag
  • help: None
• masknet:
  • bn_chan: 128
  • hid_chan: 512
  • mask_act: relu
  • n_blocks: 8
  • n_repeats: 3
  • n_src: 3
  • skip_chan: 128
• optim:
  • lr: 0.001
  • optimizer: adam
  • weight_decay: 0.0
• positional arguments:
• training:
  • batch_size: 4
  • early_stop: True
  • epochs: 200
  • half_lr: True
  • num_workers: 4

 

Results:

• si_sdr: 6.824750632456865
• si_sdr_imp: 11.234803761803752
• sdr: 7.715799858488098
• sdr_imp: 11.778681386239114
• sir: 16.442141130818637
• sir_imp: 19.527535070051055
• sar: 8.757864265661263
• sar_imp: -0.15657258049670303
• stoi: 0.7854554136619554
• stoi_imp: 0.22267957718163015

 

License notice:

This work "ConvTasNet_Libri3Mix_sepnoisy" 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_Libri3Mix_sepnoisy" is licensed under Attribution-ShareAlike 3.0 Unported by TakhirMirzaev.