elishowk's picture
Automatic correction of README.md metadata. Contact [email protected] for any question
da1de55
metadata
tags:
  - asteroid
  - audio
  - ConvTasNet
  - audio-to-audio
datasets:
  - Libri3Mix
  - sep_noisy
license: cc-by-sa-4.0

Asteroid model JorisCos/ConvTasNet_Libri3Mix_sepnoisy_8k

Description:

This model was trained by Joris Cosentino using the librimix 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
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
training:
  batch_size: 24
  early_stop: true
  epochs: 200
  half_lr: true
  num_workers: 4

Results:

On Libri3Mix min test set :

si_sdr: 5.978836560066222
si_sdr_imp: 10.388889689413096
sdr: 6.8651365291740225
sdr_imp: 10.928018056925016
sir: 14.997089638783114
sir_imp: 18.08248357801549
sar: 8.127504792061933
sar_imp: -0.7869320540959925
stoi: 0.7669414686111115
stoi_imp: 0.20416563213078837

License notice:

This work "ConvTasNet_Libri3Mix_sepnoisy_8k" 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_8k" is licensed under Attribution-ShareAlike 3.0 Unported by Joris Cosentino