Update README.md
Browse files
README.md
CHANGED
|
@@ -28,23 +28,21 @@ tags:
|
|
| 28 |
|
| 29 |
## Supported Languages
|
| 30 |
|
| 31 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 32 |
|
| 33 |
-
|
| 34 |
-
lang2id:
|
| 35 |
-
chinese: 0
|
| 36 |
-
dutch: 1
|
| 37 |
-
english: 2
|
| 38 |
-
french: 3
|
| 39 |
-
german: 4
|
| 40 |
-
italian: 5
|
| 41 |
-
japanese: 6
|
| 42 |
-
other: 7 # Vocal techniques
|
| 43 |
-
polish: 8
|
| 44 |
-
portuguese: 9
|
| 45 |
-
spanish: 10
|
| 46 |
-
```
|
| 47 |
-
Note: The "other" category includes various vocal techniques.
|
| 48 |
|
| 49 |
## Model Overview
|
| 50 |
FreeSVC leverages an enhanced VITS architecture integrated with Speaker-invariant Clustering (SPIN) and the ECAPA2 speaker encoder. This combination effectively separates speaker characteristics from linguistic content, ensuring high-quality and natural-sounding voice conversions across multiple languages.
|
|
|
|
| 28 |
|
| 29 |
## Supported Languages
|
| 30 |
|
| 31 |
+
| Language | ID | Status | Speech Data | Singing Data |
|
| 32 |
+
|------------|-----|--------------|-------------|--------------|
|
| 33 |
+
| Chinese | 0 | ✅ Full | 255h | 70h |
|
| 34 |
+
| Dutch | 1 | ✅ Full | Part of CML | - |
|
| 35 |
+
| English | 2 | ✅ Full | 921h | 47h |
|
| 36 |
+
| French | 3 | ✅ Full | Part of CML | - |
|
| 37 |
+
| German | 4 | ✅ Full | Part of CML | - |
|
| 38 |
+
| Italian | 5 | ✅ Full | Part of CML | - |
|
| 39 |
+
| Japanese | 6 | ✅ Full | 30h | - |
|
| 40 |
+
| Other* | 7 | ⚠️ Partial | - | 10h |
|
| 41 |
+
| Polish | 8 | ✅ Full | Part of CML | - |
|
| 42 |
+
| Portuguese | 9 | ✅ Full | Part of CML | - |
|
| 43 |
+
| Spanish | 10 | ✅ Full | Part of CML | - |
|
| 44 |
|
| 45 |
+
*Note: The "Other" category is used for vocal techniques without content.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 46 |
|
| 47 |
## Model Overview
|
| 48 |
FreeSVC leverages an enhanced VITS architecture integrated with Speaker-invariant Clustering (SPIN) and the ECAPA2 speaker encoder. This combination effectively separates speaker characteristics from linguistic content, ensuring high-quality and natural-sounding voice conversions across multiple languages.
|