File size: 1,524 Bytes
de14453 c0461ed 697e7ab |
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 51 52 53 54 55 56 57 58 59 |
---
library_name: transformers
tags: []
---
# Model Card for Model ID
---
language: lo
tags:
- audio
- automatic-speech-recognition
- wav2vec2
- lao
license: apache-2.0
model-index:
- name: Wav2Vec2 Lao Fine-tuned
results: []
---
# Wav2Vec2 Lao Fine-tuned
This model is a fine-tuned version of [`facebook/wav2vec2-xls-r-300m`](https://huggingface.co/facebook/wav2vec2-xls-r-300m) on Lao speech data. It was trained for automatic speech recognition (ASR) using [SiangLao](https://huggingface.co/datasets/sianglao) or similar datasets.
## Intended Use
- Lao language ASR tasks
- Research in low-resource language modeling
## Training Details
- Base model: facebook/wav2vec2-xls-r-300m
- Framework: Hugging Face Transformers
- Fine-tuned on: Lao speech dataset
- Tokenizer and processor: see [`wav2vec2-lao-processor`](https://huggingface.co/YourUsername/wav2vec2-lao-processor)
## How to Use
```python
from transformers import Wav2Vec2Processor, Wav2Vec2ForCTC
import torch
import torchaudio
processor = Wav2Vec2Processor.from_pretrained("YourUsername/wav2vec2-lao-processor")
model = Wav2Vec2ForCTC.from_pretrained("YourUsername/wav2vec2-lao-finetuned")
# Load audio
waveform, sample_rate = torchaudio.load("your_audio.wav")
# Preprocess
inputs = processor(waveform.squeeze(), sampling_rate=sample_rate, return_tensors="pt", padding=True)
with torch.no_grad():
logits = model(**inputs).logits
# Decode
predicted_ids = torch.argmax(logits, dim=-1)
transcription = processor.batch_decode(predicted_ids)
```
|