large-audio-language-models-asked-return-every-time-span CLM-0583When large audio language models are asked to return every time span matching a natural-language query, they almost always report fewer intervals than exist: across 21 systems on queries averaging about 4 ground-truth intervals, under-reporting rates run from 79% to 100% and no system exceeds 13.2% exact count accuracy, even for the models with the best overlap scores.
Capability: Understanding speech and audio
Observed on
Free-form timestamp output, 1,750 human-verified query-recording pairs (7s to 20 min), 387 queries with two or more ground-truth intervals, no oracle count or explicit enumeration .
Sources
- Validated by: Applied across 21 systems; it exposed that two weak models' apparently higher one-to-many mIoU came from emitting single long intervals that accrue intersection by accident, which mIoU alone could not reveal.. Measured over 387 multi-interval queries with mean 4.02 ground-truth intervals; average predicted intervals ranged 0.18-3.67, most systems near one. Parsing failures count as zero intervals, so some under-reporting is format failure rather than enumeration failure, and the paper says the two cannot be fully separated. [truncated]
Disagreeing is the most useful thing you can do here. Nothing on this page waits for a reviewer: the counts above move when evidence lands, and every source stays attached with who brought it.
Related claims
- For spoken language models answering questions over long dialogue histories, converting the history to transcripts beats keeping it as audio when the answer is a lexical detail (exact slot values across dense turns: mean 71.3% vs 42.6%), but loses when the answer depends on how something was said, where audio history wins for six of seven models and by a larger margin once lexical emotion cues are removed.Understanding speech and audio · reviewed by AI
- For large audio language models on multiple-choice audio reasoning benchmarks (MMAR, MMAU-mini), accuracy does not track whether the model's chain of thought is acoustically grounded or causally drives the answer: several models answer correctly from perception steps that assert unsupported labels, and for some models replacing the reasoning chain with vacuous or fabricated content leaves the final answer unchanged.Understanding speech and audio · reviewed by AI
- In a full-duplex speech-to-speech model that emits function calls on a dedicated parallel output stream, choosing the right tool from spoken, disfluent user requests is far easier than filling its arguments: tool-selection F1 reached 82.5% (above two closed realtime APIs) while argument accuracy was 42.2% and end-to-end Pass@1 33.0%, below both.Using the tools it is given · reviewed by AI
- On multi-session dialogue questions whose answers depend only on how an utterance was delivered, a cascaded ASR-then-LLM pipeline (Whisper large-v3 into a frontier model) scores below 7B audio-native models on the same clips, and below even a transcript-only baseline, because the transcription step drops delivery cues entirely and adds content errors.Understanding speech and audio · reviewed by AI
- In audio language models that feed encoder outputs into a text LLM, downstream understanding accuracy across speech, sound and music is set by how semantically text-aligned the encoder is rather than by whether its representation is discrete or continuous — reconstruction-oriented codecs with high acoustic fidelity underperform, semantically distilled tokenizers can beat their own continuous teacher, and scaling the LLM backbone does not recover the lost information.Understanding speech and audio · reviewed by AI
Notes
Drafted from the paper by a model and filed unreviewed. Visible here so it can be read, not because anyone has vouched for it: it does not move any technique's standing and does not count toward the internal scorecard. Drafted confidence: medium. Falsifier as drafted: A model evaluated under the same free-form protocol that emits roughly the correct number of intervals on recurring-event queries (count accuracy well above 13%) without an oracle count, or evidence that a plain 'enumerate all occurrences' prompt removes the under-reporting.