badrex/w2v-bert-2.0-malagasy-asr

Automatic Speech Recognition for Malagasy

Model Description 🍍

This model is a fine-tuned version of Wav2Vec2-BERT 2.0 for Malagasy automatic speech recognition (ASR). It was trained on 150 hours of transcribed Malagasy speech. The ASR model is robust and the in-domain WER is below 11.7%.

• Developed by: Badr al-Absi
• Model type: Speech Recognition (ASR)
• Language: Malagasy (mg)
• License: CC-BY-4.0
• Finetuned from: facebook/w2v-bert-2.0

Direct Use

The model can be used directly for automatic speech recognition of a Malagasy audio:

from transformers import Wav2Vec2BertProcessor, Wav2Vec2BertForCTC
import torch
import torchaudio

# load model and processor
processor = Wav2Vec2BertProcessor.from_pretrained("badrex/w2v-bert-2.0-malagasy-asr")
model = Wav2Vec2BertForCTC.from_pretrained("badrex/w2v-bert-2.0-malagasy-asr")

# load audio
audio_input, sample_rate = torchaudio.load("path/to/audio.wav")

# preprocess
inputs = processor(audio_input.squeeze(), sampling_rate=sample_rate, return_tensors="pt")

# inference
with torch.no_grad():
    logits = model(**inputs).logits

# decode
predicted_ids = torch.argmax(logits, dim=-1)
transcription = processor.batch_decode(predicted_ids)[0]
print(transcription)

Downstream Use

This model can be used as a foundation for:

• building voice assistants for Malagasy speakers
• transcription services for Malagasy content
• accessibility tools for Malagasy-speaking communities
• research in low-resource speech recognition

Model Architecture

• Base model: Wav2Vec2-BERT 2.0
• Architecture: transformer-based with convolutional feature extractor
• Parameters: ~600M (inherited from base model)
• Objective: connectionist temporal classification (CTC)

Funding

The development of this model was supported by CLEAR Global and Gates Foundation.

Citation

@misc{w2v_bert_malagasy_asr,
  author = {Badr M. Abdullah},
  title = {Adapting Wav2Vec2-BERT 2.0 for Malagasy ASR},
  year = {2025},
  publisher = {Hugging Face},
  url = {https://huggingface.co/badrex/w2v-bert-2.0-malagasy-asr}
}

Model Card Contact

For questions or issues, please contact via the Hugging Face model repository in the community discussion section.