Polyglot-Lion-1.7B: High-accuracy multilingual ASR for Singapore — English, Mandarin, Tamil & Malay

CHANGE LOG: This version was retrained on the same dataset without punctuation removal to improve the model’s ability to recognize pauses and sentence boundaries in speech.

About

Polyglot-Lion-1.7B was developed by Quy-Anh Dang and Chris Ngo at Knovel Engineering and presented in the report "Polyglot-Lion: Efficient Multilingual ASR for Singapore via Balanced Fine-Tuning of Qwen3-ASR".

The model is obtained by fine-tuning Qwen3-ASR-1.7B exclusively on publicly available speech corpora covering Singapore's four official languages. It utilizes a balanced sampling strategy that equalizes the number of training utterances per language and deliberately omits language-tag conditioning, allowing the model to learn to identify languages implicitly from audio.

Polyglot-Lion-1.7B achieves an average error rate of 14.85 — competitive with MERaLiON-2-10B-ASR (14.32), a model 6× larger and 20× faster inference.

• Parameters: 1.7B
• Languages: English, Mandarin, Tamil, Malay
• Training cost: $81 on a single NVIDIA RTX PRO 6000 (48 h)
• Inference speed: ~0.10 s/sample on RTX PRO 4500

Results

WER (%) for English, Tamil, and Malay; CER (%) for Mandarin. Lower is better. Bold = best overall.

Quick Start

Polyglot-Lion uses the qwen-asr package for inference.

# Install uv (if not already installed)
curl -LsSf https://astral.sh/uv/install.sh | sh

# Create environment and install
uv venv --python 3.12 && source .venv/bin/activate
uv pip install qwen-asr hf_transfer

Transformers

import torch
from qwen_asr import Qwen3ASRModel

model = Qwen3ASRModel.from_pretrained(
    "knoveleng/polyglot-lion-1.7b",
    dtype=torch.bfloat16,
    device_map="cuda:0",
    max_new_tokens=256,
)

results = model.transcribe(audio="path/to/audio.wav", language=None)
print(results[0].language, results[0].text)

vLLM (faster)

from qwen_asr import Qwen3ASRModel

if __name__ == "__main__":
    model = Qwen3ASRModel.LLM(
        model="knoveleng/polyglot-lion-1.7b",
        gpu_memory_utilization=0.7,
        max_new_tokens=4096,
    )
    results = model.transcribe(audio=["audio1.wav", "audio2.wav"], language=None)
    for r in results:
        print(r.language, r.text)

For batch inference, timestamps, streaming, and server deployment, see the Qwen3-ASR documentation.

Citation

@misc{dang2026polyglotlion,
    title={Polyglot-Lion: Efficient Multilingual ASR for Singapore via Balanced Fine-Tuning of Qwen3-ASR}, 
    author={Quy-Anh Dang and Chris Ngo},
    year={2026},
    eprint={2603.16184},
    archivePrefix={arXiv},
    primaryClass={cs.CL},
    url={https://arxiv.org/abs/2603.16184}, 
}