Commit
•
cc7479f
1
Parent(s):
297cd7c
Update README.md
Browse files
README.md
CHANGED
@@ -42,15 +42,16 @@ metrics:
|
|
42 |
|
43 |
<br/><br/>
|
44 |
|
45 |
-
#
|
46 |
|
47 |
-
This repository provides
|
48 |
|
49 |
-
| Release | Test-Set SI-SNRi | Test-Set SDRi |
|
50 |
|
51 |
-
|
52 |
|
53 |
-
|
|
|
|
|
54 |
|
55 |
## Install SpeechBrain
|
56 |
|
@@ -67,20 +68,13 @@ Please notice that we encourage you to read our tutorials and learn more about [
|
|
67 |
### Perform source separation on your own audio file
|
68 |
|
69 |
```python
|
|
|
|
|
70 |
|
71 |
-
|
72 |
-
|
73 |
-
import torchaudio
|
74 |
-
|
75 |
-
model = separator.from_hparams(source="speechbrain/sepformer-whamr16k", savedir='pretrained_models/sepformer-whamr16k')
|
76 |
-
|
77 |
-
# for custom file, change path
|
78 |
|
79 |
-
|
80 |
-
|
81 |
-
torchaudio.save("source1hat.wav", est_sources[:, :, 0].detach().cpu(), 16000)
|
82 |
-
|
83 |
-
torchaudio.save("source2hat.wav", est_sources[:, :, 1].detach().cpu(), 16000)
|
84 |
|
85 |
```
|
86 |
|
@@ -117,11 +111,9 @@ pip install -e .
|
|
117 |
3. Run Training:
|
118 |
|
119 |
```
|
|
|
120 |
|
121 |
-
|
122 |
-
|
123 |
-
python train.py hparams/sepformer-whamr.yaml --data_folder=your_data_folder --sample_rate=16000
|
124 |
-
|
125 |
```
|
126 |
|
127 |
You can find our training results (models, logs, etc) [here](https://drive.google.com/drive/folders/1QiQhp1vi5t4UfNpNETA48_OmPiXnUy8O?usp=sharing).
|
|
|
42 |
|
43 |
<br/><br/>
|
44 |
|
45 |
+
# SI-SNR Estiamtor
|
46 |
|
47 |
+
This repository provides the SI-SNR estimator model introduced for the REAL-M dataset.
|
48 |
|
|
|
49 |
|
50 |
+
| Release | Test-Set (WHAMR!) average l1 error |
|
51 |
|
52 |
+
|:-------------:|:--------------:|
|
53 |
+
|
54 |
+
| 18-10-21 | 1.7 dB |
|
55 |
|
56 |
## Install SpeechBrain
|
57 |
|
|
|
68 |
### Perform source separation on your own audio file
|
69 |
|
70 |
```python
|
71 |
+
model = separator.from_hparams(source="speechbrain/sepformer-whamr", savedir='pretrained_models/sepformer-whamr2')
|
72 |
+
est_sources = model.separate_file(path='speechbrain/sepformer-wsj02mix/test_mixture.wav')
|
73 |
|
74 |
+
snr_est_model = snrest.from_hparams(source="speechbrain/REAL-M-sisnr-estimator-main")
|
|
|
|
|
|
|
|
|
|
|
|
|
75 |
|
76 |
+
mix, fs = torchaudio.load('test_mixture.wav')
|
77 |
+
snrhat = snr_est_model.estimate_batch(mix, est_sources)
|
|
|
|
|
|
|
78 |
|
79 |
```
|
80 |
|
|
|
111 |
3. Run Training:
|
112 |
|
113 |
```
|
114 |
+
cd recipes/REAL-M/sisnr-estimation
|
115 |
|
116 |
+
python train.py hparams/pool_sisnrestimator.yaml --data_folder /yourLibri2Mixpath --base_folder_dm /yourLibriSpeechpath --rir_path /yourpathforwhamrRIRs --dynamic_mixing True --use_whamr_train True --whamr_data_folder /yourpath/whamr --base_folder_dm_whamr /yourpath/wsj0-processed/si_tr_s
|
|
|
|
|
|
|
117 |
```
|
118 |
|
119 |
You can find our training results (models, logs, etc) [here](https://drive.google.com/drive/folders/1QiQhp1vi5t4UfNpNETA48_OmPiXnUy8O?usp=sharing).
|