File size: 1,054 Bytes
a37574e 3689890 8e0de45 db6d8fd 8e0de45 db6d8fd 8e0de45 db6d8fd 8e0de45 |
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 |
---
license: apache-2.0
base_model:
- neuphonic/neucodec
pipeline_tag: audio-to-audio
---
# NeuCodec Onnx Decoder
This is an onnx-compiled version of the decoder of [NeuCodec](https://huggingface.co/neuphonic/neucodec).
It's main use case is providing a low footprint decoder for on-device TTS.
## Usage
```python
import librosa
import torch
import soundfile as sf
from torchaudio import transforms as T
from neucodec import NeuCodec, NeuCodecOnnxDecoder
model = NeuCodec.from_pretrained("neuphonic/neucodec")
model.eval()
compiled_model = NeuCodecOnnxDecoder.from_pretrained("neuphonic/neucodec-onnx-decoder")
y, sr = torchaudio.load(librosa.ex("libri1"))
if sr != 16_000:
y = T.Resample(sr, 16_000)(y)[None, ...] # (B, 1, T_16)
with torch.no_grad():
fsq_codes = model.encode_code(y)
# fsq_codes = model.encode_code(librosa.ex("libri1")) # or directly pass your filepath!
print(f"Codes shape: {fsq_codes.shape}")
recon = compiled_model.decode_code(fsq_codes) # (B, 1, T_24)
sf.write("reconstructed.wav", recon, 24_000)
``` |