模型:
JorisCos/DCUNet_Libri1Mix_enhsingle_16k
任务:
音频到音频许可:
cc-by-sa-4.0Description:
This model was trained by Joris Cosentino using the librimix 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-360 valid_dir: data/wav16k/min/dev filterbank: stft_n_filters: 1024 stft_kernel_size: 1024 stft_stride: 256 masknet: architecture: Large-DCUNet-20 fix_length_mode: pad n_src: 1 optim: lr: 0.001 optimizer: adam weight_decay: 1.0e-05 training: batch_size: 2 early_stop: true epochs: 200 gradient_clipping: 5 half_lr: true num_workers: 4
Results:
On Libri1Mix min test set :
si_sdr: 13.154035391645971 si_sdr_imp: 9.704254085786271 sdr: 13.568058873121435 sdr_imp: 10.065396073908367 sar: 13.568058873121435 sar_imp: 10.065396073908367 stoi: 0.9199373340235417 stoi_imp: 0.12401751048300132
License notice:
This work "DCUNet_Libri1Mix_enhsignle_16k" 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). "DCUNet_Libri1Mix_enhsignle_16k" is licensed under Attribution-ShareAlike 3.0 Unported by Joris Cosentino