69 lines
2.0 KiB
Markdown
69 lines
2.0 KiB
Markdown
---
|
|
library_name: transformers
|
|
tags:
|
|
- speech
|
|
- audio
|
|
- wav2vec2
|
|
- automatic-speech-recognition
|
|
pipeline_tag: automatic-speech-recognition
|
|
---
|
|
|
|
# omniASR-CTC-1B-v2
|
|
|
|
Wav2Vec2 CTC ASR model (v2) converted from the [OmniLingual](https://github.com/facebookresearch/omnilingual-asr) fairseq2 checkpoint `omniASR_CTC_1B_v2`.
|
|
|
|
This model outputs CTC logits over a SentencePiece vocabulary and can transcribe speech in multiple languages.
|
|
|
|
# Code Base
|
|
|
|
The code base for the conversion can be found [here](https://github.com/ahmedadelattia/omnilingual_to_hf). I was only able to convert the 300M and 1B models due to GPU limitations. Contributions are welcome.
|
|
|
|
## Model details
|
|
|
|
| Property | Value |
|
|
|---|---|
|
|
| HF class | `Wav2Vec2ForCTC` |
|
|
| Encoder layers | 48 |
|
|
| Hidden size | 1280 |
|
|
| Attention heads | 16 |
|
|
| FFN intermediate | 5120 |
|
|
| Vocabulary size | 10288 |
|
|
| Source framework | fairseq2 |
|
|
| Source card | `omniASR_CTC_1B_v2` |
|
|
| Parity verification | ✅ Verified |
|
|
|
|
|
|
Numerical parity against the original fairseq2 checkpoint has been confirmed: outputs match to within `atol=1e-4` on a held-out audio sample.
|
|
|
|
Sample transcriptions on the held-out audio clip:
|
|
|
|
| Model | Transcript |
|
|
|---|---|
|
|
| fairseq2 (source) | `concord returned to its place amidst the tents` |
|
|
| HuggingFace (this repo) | `concord returned to its place amidst the tents` |
|
|
|
|
## Usage
|
|
|
|
```python
|
|
from transformers import Wav2Vec2ForCTC, AutoProcessor
|
|
import torch, torchaudio
|
|
|
|
processor = AutoProcessor.from_pretrained("aadel4/omniASR-CTC-1B-v2")
|
|
model = Wav2Vec2ForCTC.from_pretrained("aadel4/omniASR-CTC-1B-v2")
|
|
model.eval()
|
|
|
|
waveform, sr = torchaudio.load("audio.wav")
|
|
if sr != 16_000:
|
|
waveform = torchaudio.functional.resample(waveform, sr, 16_000)
|
|
|
|
inputs = processor(
|
|
waveform.squeeze().numpy(), sampling_rate=16_000, return_tensors="pt"
|
|
)
|
|
with torch.no_grad():
|
|
logits = model(**inputs).logits # (1, T, vocab)
|
|
|
|
pred_ids = torch.argmax(logits, dim=-1)
|
|
transcript = processor.decode(pred_ids[0])
|
|
print(transcript)
|
|
```
|