T2AV-Compass: A Unified Benchmark for Text-to-Audio-Video Generation

T2AV-Compass is a benchmark dataset created by NJU-LINK for evaluating Text-to-Audio-Video (T2AV) generation models. It targets unimodal quality, cross-modal alignment, complex instruction following, and perceptual realism. The dataset was last updated on December 25, 2025.

Use Cases

• Benchmarking video quality in T2AV models based on the unimodal evaluation target
• Assessing audio-video synchronization based on the cross-modal alignment target
• Testing model adherence to complex instructions based on the instruction-following target
• Evaluating perceptual realism based on the physical and common-sense constraints target

Strengths

• Explicitly targets multiple evaluation dimensions: unimodal quality, cross-modal alignment, instruction following, and realism
• Created by a known research group (NJU-LINK)
• Last updated on 2025-12-25, indicating recent maintenance

Limitations

• Description metadata is limited; actual data quality requires manual inspection after download
• Column-level documentation is absent; field semantics must be inferred after download
• Row count, file formats, and license are unknown, which may limit suitability assessment

Provenance

Source

NJU-LINK

Collection Method

Likely constructed for benchmarking purposes; specific gathering method is not detailed in the provided description.

Time Range

null

Freshness

Last updated 2025-12-25 16:36:31

Geography

null