David-A-Amoo/Qwen3-ASR-O.6B-Yoruba

Qwen3-ASR Fine-Tuned on Yoruba

This model is a fine-tuned version of Qwen/Qwen3-ASR-0.6B trained on 85.49 hours of Yoruba speech from a cleaned version of the NaijaVoices dataset.

Currently, only evaluation/testing checkpoints are available. Works extremely well for some samples, fails for others, good for short clips

Usage

import torch
from qwen_asr import Qwen3ASRModel

model = Qwen3ASRModel.from_pretrained("David-A-Amoo/qwen3-asr-yoruba")

results = model.transcribe(audio="audio.wav")

print(results[0].language)
print(results[0].text)

cc-by-nc-sa-4.0 license inheriting naijavoices dataset licensing.