This model was trained by wyz based on the universalsev1 recipe in espnet. More information can be found at
Source du modèle
Extrait de la source
This model was trained by wyz based on the universalsev1 recipe in espnet. More information can be found at
Sources
1 sourceVérifié 18 sept.
Artefacts du modèle
1 artefactexp_vctk_dns20/enh_train_enh_bsrnn_tiny_noncausal_raw/97epoch.pth
pth · 5,86 MB · SHA-256 67c933000f29…1219 · Hugging Face
TéléchargerExtraits de sources
2 extraits--- tags: - espnet - audio - audio-to-audio language: en datasets: - VCTK_DEMAND - DNS2020 license: cc-by-4.0 --- ## ESPnet2 ENH model ### `wyz/vctk_dns2020_bsrnn_tiny_noncausal` This model was trained by wyz based on the universal_se_v1 recipe in [espnet](https://github.com/espnet/espnet/). More information can be found at https://github.com/Emrys365/se-scaling. ### Demo: How to use in ESPnet2 Follow the [ESPnet installation instructions](https://espnet.github.io/espnet/installation.html) if you haven't done that already. To use the model in the Python interface, you could use the following code: ```python import soundfile as sf from espnet2.bin.enh_inference import SeparateSpeech # For model downloading + loading model = SeparateSpeech.from_pretrained( model_tag="wyz/vctk_dns2020_bsrnn_tiny_noncausal", normalize_output_wav=True, device="cuda", ) # For loading a downloaded model # model = SeparateSpeech( # train_config="exp_vctk_dns20/enh_train_enh_bsrnn_tiny_noncausal_raw/config.yaml", # model_file="exp_vctk_dns20/enh_train_enh_bsrnn_tiny_noncausal_raw/xxxx.pth", # normalize_output_wav=True, # device="cuda", # ) audio, fs = sf.read("/path/to/noisy/utt1.flac") enhanced = model(audio[None, :], fs=fs)[0] ``` <!-- Generated by ./scripts/utils/show_enh_score.sh --> # RESULTS ## Environments - date: `Tue Feb 27 22:24:50 EST 2024` - python version: `3.8.16 (default, Mar 2 2023, 03:21:46) [GCC 11.2.0]` - espnet version: `espnet 202304` - pytorch version: `pytorch 2.0.1+cu118` - Git hash: `443028662106472c60fe8bd892cb277e5b488651` - Commit date: `Thu May 11 03:32:59 2023 +0000` ## enhanced_test_16k |dataset|PESQ_WB|STOI|SAR|SDR|SIR|SI_SNR|OVRL|SIG|BAK|P808_MOS| |---|---|---|---|---|---|---|---|---|---|---| |chime4_et05_real_isolated_6ch_track|1.17|53.08|-2.74|-2.74|0.00|-30.98|2.85|3.18|3.81|3.65| |chime4_et05_simu_isolated_6ch_track|1.50|82.97|8.37|8.37|0.00|2.38|2.74|3.04|3.88|3.32| |dns20_tt_synthetic_no_reverb|3.12|97.48|19.04|19.04|0.00|19.01|3.30|3.56|4.08|4.02| |reverb_et_real_8ch_multich|1.81|85.52|11.17|11.17|0.00|7.26|2.76|3.21|3.57|3.60| |reverb_et_simu_8ch_multich|1.62|85.35|9.19|9.19|0.00|-10.63|2.61|3.17|3.19|3.58| |whamr_tt_mix_single_reverb_max_16k|1.49|84.18|7.41|7.41|0.00|2.22|2.82|3.20|3.74|3.68| ## enhanced_test_48k |dataset|STOI|SAR|SDR|SIR|SI_SNR|OVRL|SIG|BAK|P808_MOS| |---|---|---|---|---|---|---|---|---|---| |vctk_noisy_...
Source context: 1 downloads · 0 likes · Pipeline audio-to-audio · Library espnet · Repo wyz/vctk_dns2020_bsrnn_tiny_noncausal