harshit345
/
xlsr-wav2vec-speech-emotion-recognition
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

Update README.md

This commit is contained in:
harshit katyal 2021-12-12 20:43:57 +00:00 committed by huggingface-web
parent f7cac3045a
commit 2c372c0aa6
1 changed files with 27 additions and 48 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

75
README.md
View File

@ -1,25 +1,24 @@
---
language: en
datasets: Toronto emotional speech set (TESS)(https://www.kaggle.com/ejlok1/toronto-emotional-speech-set-tess)
datasets:
- aesdd
tags:
- audio
- automatic-speech-recognition
- audio-classification
- speech
- speech-emotion-recognition
license: apache-2.0
---
# Emotion Recognition in Speech using Wav2Vec 2.0
## How to use
### Requirements
```bash
~~~
# requirement packages
!pip install git+https://github.com/huggingface/datasets.git
!pip install git+https://github.com/huggingface/transformers.git
!pip install torchaudio
!pip install librosa
```
### Prediction
```python
!git clone https://github.com/m3hrdadfi/soxan
cd soxan
~~~
# prediction
~~~
import torch
import torch.nn as nn
import torch.nn.functional as F
@ -29,16 +28,16 @@ import librosa
import IPython.display as ipd
import numpy as np
import pandas as pd
```
```python
~~~
~~~
device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
model_name_or_path = "harshit345/xlsr-wav2vec-speech-emotion-recognition"
model_name_or_path = "Bagus/wav2vec2-xlsr-greek-speech-emotion-recognition"
config = AutoConfig.from_pretrained(model_name_or_path)
feature_extractor = Wav2Vec2FeatureExtractor.from_pretrained(model_name_or_path)
sampling_rate = feature_extractor.sampling_rate
model = Wav2Vec2ForSpeechClassification.from_pretrained(model_name_or_path).to(device)
```
```python
~~~
~~~
def speech_file_to_array_fn(path, sampling_rate):
speech_array, _sampling_rate = torchaudio.load(path)
resampler = torchaudio.transforms.Resample(_sampling_rate)
@ -53,37 +52,17 @@ def predict(path, sampling_rate):
scores = F.softmax(logits, dim=1).detach().cpu().numpy()[0]
outputs = [{"Emotion": config.id2label[i], "Score": f"{round(score * 100, 3):.1f}%"} for i, score in enumerate(scores)]
return outputs
```
```python
path = "/path/to/disgust.wav"
~~~
# prediction
~~~
# path for a sample
path = '/data/jtes_v1.1/wav/f01/ang/f01_ang_01.wav'
outputs = predict(path, sampling_rate)
```
```bash
[
{'Emotion': 'anger', 'Score': '12.2%'},
{'Emotion': 'disgust', 'Score': '78.8%'},
{'Emotion': 'fear', 'Score': '7.2%'},
{'Emotion': 'happiness', 'Score': '1.3%'},
{'Emotion': 'sadness', 'Score': '1.5%'}
]
```
## Evaluation
The following tables summarize the scores obtained by model overall and per each class.
| Emotions | precision | recall | f1-score | accuracy |
|-----------|-----------|--------|----------|----------|
| anger | 0.82 | 1.00 | 0.81 | |
| disgust | 0.85 | 0.96 | 0.85 | |
| fear | 0.78 | 0.88 | 0.80 | |
| happiness | 0.84 | 0.71 | 0.78 | |
| sadness | 0.86 | 1.00 | 0.79 | |
| | | | Overall | 0.806 |
##
Colab Notebook
https://colab.research.google.com/drive/1aPPb_ZVS5dlFVZySly8Q80a44La1XjJu?usp=sharing
~~~
~~~
[{'Emotion': 'anger', 'Score': '98.3%'},
{'Emotion': 'disgust', 'Score': '0.0%'},
{'Emotion': 'fear', 'Score': '0.4%'},
{'Emotion': 'happiness', 'Score': '0.7%'},
{'Emotion': 'sadness', 'Score': '0.5%'}]
~~~