File size: 1,156 Bytes
f150b2c
 
 
 
 
f0db202
3ddd487
ed8beae
 
 
3ddd487
ed8beae
f0db202
 
 
 
 
 
ed8beae
f0db202
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
e21695a
f0db202
 
 
 
 
 
 
 
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
---
license: cc-by-4.0
tags:
- Speech tokenizer
---
# Getting Started with XCodec2 on Hugging Face
XCodec2 is a speech tokenizer that offers the following key features:

1. **Single Vector Quantization**
2. **50 Tokens per Second**
3. **Multilingual Speech Semantic Support and High-Quality Speech Reconstruction**


To use `xcodec2`, ensure you have it installed. You can install it using the following command:

```bash
conda create -n xcodec2 python=3.9
conda activate xcodec2
pip install xcodec2==0.1.1
```
Then,
```python
import torch
import soundfile as sf
from transformers import AutoConfig

 
from xcodec2.modeling_xcodec2 import XCodec2Model
 
model_path = "HKUST-Audio/xcodec2"  
 
model = XCodec2Model.from_pretrained(model_path)
model.eval().cuda()   

 
wav, sr = sf.read("test.wav")   
wav_tensor = torch.from_numpy(wav).float().unsqueeze(0)  

 
with torch.no_grad():
   # only 16khz speech 
    vq_code = model.encode_code(input_waveform=wav_tensor)
    print("Code:", vq_code )   

    recon_wav = model.decode_code(vq_code).cpu()

 
sf.write("reconstructed.wav", recon_wav[0, 0, :].numpy(), sr)
print("Done! Check reconstructed.wav")
```