MMAU#

Overview#

MMAU (Massive Multitask Audio Understanding) is a comprehensive benchmark for evaluating audio understanding capabilities of multimodal large language models across diverse audio tasks.

Task Description#

Task Type: Audio Understanding (Multiple Choice)
Input: Audio recordings with multiple-choice questions
Output: Correct answer choice (A/B/C/D)
Categories: Speech, Sound, Music

Key Features#

Large-scale audio understanding benchmark
Covers multiple audio domains (speech, environmental sounds, music)
Multiple-choice format with 4 options
Includes both mini and full test sets
Per-category accuracy reporting

Evaluation Notes#

Default configuration uses test_mini split
Primary metric: Accuracy (exact match on predicted letter)
Reports overall accuracy and per-task-category accuracy
Prompt includes chain-of-thought instruction

Properties#

Property	Value
Benchmark Name	`mmau`
Dataset ID	lmms-lab/mmau
Paper	N/A
Tags	`Audio`, `MCQ`
Metrics	`acc`
Default Shots	0-shot
Evaluation Split	`test_mini`

Data Statistics#

Statistics not available.

Sample Example#

Sample example not available.

Prompt Template#

Prompt Template:

Answer the following multiple choice question based on the audio content. The last line of your response should be of the following format: 'ANSWER: [LETTER]' (without quotes) where [LETTER] is one of {letters}. Think step by step before answering.

{question}

{choices}

Usage#

Using CLI#

evalscope eval \
    --model YOUR_MODEL \
    --api-url OPENAI_API_COMPAT_URL \
    --api-key EMPTY_TOKEN \
    --datasets mmau \
    --limit 10  # Remove this line for formal evaluation

Using Python#

from evalscope import run_task
from evalscope.config import TaskConfig

task_cfg = TaskConfig(
    model='YOUR_MODEL',
    api_url='OPENAI_API_COMPAT_URL',
    api_key='EMPTY_TOKEN',
    datasets=['mmau'],
    limit=10,  # Remove this line for formal evaluation
)

run_task(task_cfg=task_cfg)