CAST-Bench: Tracing Every Sound to Its Source in Multi-Shot Audio-Video Generation

Existing benchmarks for joint text-to-audio-video (T2AV) generation evaluate perceptual quality and audiovisual alignment, but balanced sound-source sam- pling and separate reporting by shot structure remain limited. This makes it dif- ficult to determine whether a specific sound is missing, attributed to the wrong source, or lost across shot transitions. To address this, we introduce CAST- BENCH, a benchmark designed to trace every generated sound back to its in- tended source in multi-shot audio-video generation. The core idea is to formulate T2AV correctness as an audiovisual contract in which each requested sound must be emitted by the right source, synchronized with the corresponding visual ac- tion, and carry the right semantic content. Concretely, the benchmark organizes 900 prompts into 9 sound-source categories, with 100 prompts per category, span- ning speech, animals, instruments, ambient sounds, and other emitter types. The prompts include 439 single-shot and 461 multi-shot narratives, each derived from a structured specification that links every audio event to its emitter, subject, and temporal slot. The benchmark provides an 18-metric evaluation suite covering three dimensions: Visual Performance, Content Consistency, and Audio & Se- mantic Alignment. We further benchmark 12 representative generators and find that visual quality is nearly saturated across leading systems, yet audio seman- tic scores still differ markedly. Across the nine shared models, introducing shot transitions lowers the mean by 10.71 points and changes model rankings. Multi- person speech, consistent speaker identity across shots, and cross-shot event re- tention remain the main unresolved challenges. All prompts and evaluation code will be released upon publication.