metadata

language: lo
license: apache-2.0
tags:
  - automatic-speech-recognition
  - speech
  - audio
  - lao
  - wav2vec2
  - xls-r
datasets:
  - h3llohihi/lao-asr-thesis-dataset
library_name: transformers
pipeline_tag: automatic-speech-recognition
metrics:
  - cer
base_model:
  - facebook/wav2vec2-xls-r-300m

XLS-R Lao ASR

Fine-tuned XLS-R-300M model for Lao automatic speech recognition.

Model Performance

Test CER: 15.14%
Training Time: 2.1 hours
Dialects: Central, Northern, Southern Lao

Usage

from transformers import Wav2Vec2ForCTC, Wav2Vec2Processor
import torch
import librosa

# Load model and processor
model = Wav2Vec2ForCTC.from_pretrained("h3llohihi/xls-r-lao-asr")
processor = Wav2Vec2Processor.from_pretrained("h3llohihi/xls-r-lao-asr")

# Load audio
audio, sr = librosa.load("audio.wav", sr=16000)

# Process audio
inputs = processor(audio, sampling_rate=16000, return_tensors="pt")

# Generate prediction
with torch.no_grad():
    logits = model(**inputs).logits
    predicted_ids = torch.argmax(logits, dim=-1)
    transcription = processor.batch_decode(predicted_ids)[0]

print(transcription)

Citation

@thesis{naovalath2025lao,
  title={Lao Automatic Speech Recognition using Transfer Learning},
  author={Souphaxay Naovalath and Sounmy Chanthavong},
  advisor={Dr. Somsack Inthasone},
  school={National University of Laos, Faculty of Natural Sciences, Computer Science Department},
  year={2025}
}