KELONMYOSA
commited on
Commit
•
7735ba6
1
Parent(s):
441e11d
Update README.md
Browse files
README.md
CHANGED
@@ -5,4 +5,21 @@ datasets:
|
|
5 |
language:
|
6 |
- ru
|
7 |
pipeline_tag: audio-classification
|
8 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
5 |
language:
|
6 |
- ru
|
7 |
pipeline_tag: audio-classification
|
8 |
+
metrics:
|
9 |
+
- accuracy
|
10 |
+
---
|
11 |
+
|
12 |
+
# Speech Emotion Recognition
|
13 |
+
|
14 |
+
The model is a fine-tuned version of [facebook/wav2vec2-xls-r-300m](https://huggingface.co/facebook/wav2vec2-xls-r-300m) for a Speech Emotion Recognition (SER) task.
|
15 |
+
|
16 |
+
The dataset used to fine-tune the original pre-trained model is the [DUSHA dataset](https://huggingface.co/datasets/KELONMYOSA/dusha_emotion_audio). The dataset consists of about 125 000 audio recordings in Russian with four basic emotions that usually appear in a dialog with a virtual assistant: Happiness (Positive), Sadness, Anger and Neutral emotion.
|
17 |
+
|
18 |
+
```python
|
19 |
+
emotions = ['neutral', 'positive', 'angry', 'sad', 'other']
|
20 |
+
```
|
21 |
+
|
22 |
+
It achieves the following results:
|
23 |
+
- Training Loss: 0.528700
|
24 |
+
- Validation Loss: 0.349617
|
25 |
+
- Accuracy: 0.901369
|