mispeech / r1-aqa

R1-AQA --- Reinforcement Learning Outperforms Supervised Fine-Tuning: A Case Study on Audio Question Answering

Introduction

R1-AQA is a audio question answering (AQA) model based on Qwen2-Audio-7B-Instruct , optimized through reinforcement learning using the group relative policy optimization (GRPO) algorithm. This implementation has achieved state-of-the-art performance on MMAU Test-mini benchmark with only 38k post-training samples. For more details, please refer to our Github and Technical Report .

Table: Accuracies (%) on MMAU Test-mini benchmark

Notes:

* The data are sourced from the MMAU official website: https://sakshi113.github.io/mmau_homepage/
[1] Xie, Zhifei, et al. "Audio-Reasoner: Improving Reasoning Capability in Large Audio Language Models." arXiv preprint arXiv:2503.02318 (2025).
[2] Ma, Ziyang, et al. "Audio-CoT: Exploring Chain-of-Thought Reasoning in Large Audio Language Model." arXiv preprint arXiv:2501.07246 (2025).

Inference

import torch
import torchaudio
from transformers import Qwen2AudioForConditionalGeneration, AutoProcessor

# Load model
model_name = "mispeech/r1-aqa"
processor = AutoProcessor.from_pretrained(model_name)
model = Qwen2AudioForConditionalGeneration.from_pretrained(model_name, torch_dtype=torch.bfloat16, device_map="auto")

# Load example audio
wav_path = "test-mini-audios/3fe64f3d-282c-4bc8-a753-68f8f6c35652.wav"  # from MMAU dataset
waveform, sampling_rate = torchaudio.load(wav_path)
assert sampling_rate == 16000
audios = [waveform.numpy()]

# Make prompt text
question = "Based on the given audio, identify the source of the speaking voice."
options = ["Man", "Woman", "Child", "Robot"]
prompt = f"{question} Please choose the answer from the following options: {str(options)}. Output the final answer in <answer> </answer>."
message = [
    {"role": "user", "content": [
        {"type": "audio", "audio_url": wav_path},
        {"type": "text", "text": prompt}
    ]}
]
texts = processor.apply_chat_template(message, add_generation_prompt=True, tokenize=False)

# Process
inputs = processor(text=texts, audios=audios, sampling_rate=16000, return_tensors="pt", padding=True).to(model.device)
generated_ids = model.generate(**inputs, max_new_tokens=256)
generated_ids = generated_ids[:, inputs.input_ids.size(1):]
response = processor.batch_decode(generated_ids, skip_special_tokens=True, clean_up_tokenization_spaces=False)

print(response)