Update README.md
Browse files
README.md
CHANGED
|
@@ -108,7 +108,20 @@ The embedding layer was initialized with the number of catalan speakers (47) and
|
|
| 108 |
|
| 109 |
### Training Hyperparameters
|
| 110 |
|
| 111 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 112 |
|
| 113 |
## Evaluation
|
| 114 |
|
|
|
|
| 108 |
|
| 109 |
### Training Hyperparameters
|
| 110 |
|
| 111 |
+
batch size: 32 (x2 GPUs)
|
| 112 |
+
learning rate: 1e-4
|
| 113 |
+
number of speakers: 47
|
| 114 |
+
n_fft: 1024
|
| 115 |
+
n_feats: 80
|
| 116 |
+
sample_rate: 22050
|
| 117 |
+
hop_length: 256
|
| 118 |
+
win_length: 1024
|
| 119 |
+
f_min: 0
|
| 120 |
+
f_max: 8000
|
| 121 |
+
data_statistics:
|
| 122 |
+
mel_mean: -6.578195
|
| 123 |
+
mel_std: 2.538758
|
| 124 |
+
number of samples: 13340
|
| 125 |
|
| 126 |
## Evaluation
|
| 127 |
|