Update README.md
Browse files
README.md
CHANGED
|
@@ -1,11 +1,19 @@
|
|
| 1 |
-
---
|
| 2 |
-
language:
|
| 3 |
-
- en
|
| 4 |
-
---
|
| 5 |
|
| 6 |
-
|
| 7 |
|
| 8 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 9 |
import logging
|
| 10 |
import pathlib
|
| 11 |
import re
|
|
@@ -55,7 +63,6 @@ if version.parse(torch.__version__) >= version.parse("1.6"):
|
|
| 55 |
|
| 56 |
logger = logging.getLogger(__name__)
|
| 57 |
|
| 58 |
-
|
| 59 |
@dataclass
|
| 60 |
class TrainingArguments(TrainingArguments):
|
| 61 |
output_dir: str = field(
|
|
@@ -248,4 +255,4 @@ def exeute_angry_predict(audio_path):
|
|
| 248 |
if __name__ == "__main__":
|
| 249 |
audio_path = 'audio.mp3'
|
| 250 |
exeute_angry_predict(audio_path)
|
| 251 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
|
| 2 |
+
# wav2vec2_angry_emotion 模型介绍
|
| 3 |
|
| 4 |
+
## Highlight
|
| 5 |
+
- 该模型是基于wav2vec2-base，用iemocap和Emotional Speech Dataset (ESD)数据进行微调的，故支持中文、英文音频。
|
| 6 |
+
- 准确率高达92.9%
|
| 7 |
+
- model card 展示的只是部分预测源码，详细见Files and Versions
|
| 8 |
+
- 该模型可对愤怒情绪进行预测
|
| 9 |
+
|
| 10 |
+
|
| 11 |
+
|
| 12 |
+
|
| 13 |
+
|
| 14 |
+
|
| 15 |
+
#### Some details are as follows
|
| 16 |
+
```python
|
| 17 |
import logging
|
| 18 |
import pathlib
|
| 19 |
import re
|
|
|
|
| 63 |
|
| 64 |
logger = logging.getLogger(__name__)
|
| 65 |
|
|
|
|
| 66 |
@dataclass
|
| 67 |
class TrainingArguments(TrainingArguments):
|
| 68 |
output_dir: str = field(
|
|
|
|
| 255 |
if __name__ == "__main__":
|
| 256 |
audio_path = 'audio.mp3'
|
| 257 |
exeute_angry_predict(audio_path)
|
| 258 |
+
```
|