Pith Number

pith:TYNRTIPQ

pith:2026:TYNRTIPQHLSCW7MVJCWERHQVGY

not attested not anchored not stored refs resolved

Can Large Audio Language Models Ignore Multilingual Distractors? An Evaluation of Their Selective Auditory Attention Capabilities

Heejoon Koo

Large audio language models lose selective attention to English targets when multilingual distractors appear at low signal-to-noise ratios.

arxiv:2605.17225 v1 · 2026-05-17 · eess.AS

Open paper page JSON Open Graph Bundle Merged state Verified badge What is a Pith Number?

Add to your LaTeX paper

\usepackage{pith}
\pithnumber{TYNRTIPQHLSCW7MVJCWERHQVGY}

Prints a linked badge after your title and injects PDF metadata. Compiles on arXiv. Learn more · Embed verified badge

Record completeness

1 Bitcoin timestamp

2 Internet Archive

3 Author claim open · sign in to claim

4 Citations open

5 Replications open

✓ Portable graph bundle live · download bundle · merged state

The bundle contains the canonical record plus signed events. A mirror can host it anywhere and recompute the same current state with the deterministic merge algorithm.

Claims

C1strongest claim

strong single performance does not ensure robust selective auditory attention: cocktail party accuracy degrades under severe SNRs, and errors are dominated by distractor-grounded source confusion. In addition, separation reduces acoustic overlap but leaves source attribution unresolved, often yielding confident wrong-stream answers.

C2weakest assumption

The MUSA benchmark pairs English targets with semantically plausible distractors under controlled SNRs in single, two-stage, and end-to-end settings, assuming this construction accurately measures selective auditory attention without major confounding factors from the specific dialogue content or model training data.

C3one line summary

Introduces the MUSA benchmark and evaluates LALMs showing that strong single-speaker performance fails to ensure robust selective attention under multilingual interference, with errors from source confusion and unresolved attribution after separation.

References

36 extracted · 36 resolved · 7 Pith anchors

[1] Situational awareness , pages= 2017

[2] Journal of the acoustical society of America , volume=

[3] Proceedings of the 63rd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers) , pages=

[4] arXiv preprint arXiv:2501.04962 , year=

[5] Audiotrust: Benchmarking the multifaceted trustworthiness of audio large language models

Receipt and verification

First computed	2026-05-20T00:03:46.147257Z
Builder	pith-number-builder-2026-05-17-v1
Signature	Pith Ed25519 (`pith-v1-2026-05`) · public key
Schema	pith-number/v1.0

Canonical hash

9e1b19a1f03ae42b7d9548ac489e15362e2d0d88d52d2011ae1bed8e54eda4eb

Aliases

arxiv: 2605.17225 · arxiv_version: 2605.17225v1 · doi: 10.48550/arxiv.2605.17225 · pith_short_12: TYNRTIPQHLSC · pith_short_16: TYNRTIPQHLSCW7MV · pith_short_8: TYNRTIPQ

Agent API

Resolver JSON Graph JSON Events JSON Schema Signing key

Verify this Pith Number yourself

curl -sH 'Accept: application/ld+json' https://pith.science/pith/TYNRTIPQHLSCW7MVJCWERHQVGY \
  | jq -c '.canonical_record' \
  | python3 -c "import sys,json,hashlib; b=json.dumps(json.loads(sys.stdin.read()), sort_keys=True, separators=(',',':'), ensure_ascii=False).encode(); print(hashlib.sha256(b).hexdigest())"
# expect: 9e1b19a1f03ae42b7d9548ac489e15362e2d0d88d52d2011ae1bed8e54eda4eb

Canonical record JSON

{
  "metadata": {
    "abstract_canon_sha256": "d2cee1fd02cd40e01e50d8fb6b2439be51edf9592182a086912856951b922371",
    "cross_cats_sorted": [],
    "license": "http://creativecommons.org/licenses/by-nc-nd/4.0/",
    "primary_cat": "eess.AS",
    "submitted_at": "2026-05-17T02:13:58Z",
    "title_canon_sha256": "91eca395424920448f8f8f0f2ed0fd400187c916b1602de2e5dc94304d7094b6"
  },
  "schema_version": "1.0",
  "source": {
    "id": "2605.17225",
    "kind": "arxiv",
    "version": 1
  }
}