0 kHz
Annotated moments
Select an interval to play just that target. Timestamps use the original recording’s time origin.
Hearing what happens is only half the story. TAG-Bench asks when it happens—and whether you can find every occurrence.
Alibaba Token Foundry · TAG-Bench
One minute of audio. Eighteen moments to find.
From a fraction of a second to twenty minutes. Explore the task mix, recording lengths, and how much of each recording contains the target.
Choose a duration range to filter all dataset charts.
Counts follow the original annotated intervals, before merging.
Coverage = annotated target time ÷ recording duration, using the union of intervals within the recording.
| ID | Task | Duration | Targets | Target coverage | Audio example |
|---|
These recordings contain the exact same 60-second audio segment. The target stays the same. The surrounding context grows.
Each timeline spans its full recording. The shaded region marks the embedded 60-second segment.
Sixteen real examples, two per task. Listen to complete recordings, jump to annotated moments, and explore the waveform or spectrogram.
Select an interval to play just that target. Timestamps use the original recording’s time origin.
TAG-Bench evaluates temporal audio grounding: given a recording and a natural-language query, return all matching time intervals. Accurate boundaries and complete occurrence recovery are both essential.
Read the paper ↗TAG-Bench: Benchmarking Temporal Audio Grounding in Large Audio Language Models