{
  "absUrl": "https://arxiv.org/abs/2609.02940",
  "arxivId": "2609.02940",
  "arxivVersion": null,
  "arxivVersionedId": null,
  "authors": [
    {
      "affiliations": [
        "机构信息未在 arXiv HTML 中可靠披露"
      ],
      "name": "Chan-Jan Hsu"
    },
    {
      "affiliations": [
        "机构信息未在 arXiv HTML 中可靠披露"
      ],
      "name": "Jaeyeon Kim"
    },
    {
      "affiliations": [
        "NVIDIA"
      ],
      "name": "Chao-Han Huck Yang"
    },
    {
      "affiliations": [
        "机构信息未在 arXiv HTML 中可靠披露"
      ],
      "name": "Shinji Watanabe"
    },
    {
      "affiliations": [
        "机构信息未在 arXiv HTML 中可靠披露"
      ],
      "name": "Hung-yi Lee"
    },
    {
      "affiliations": [
        "机构信息未在 arXiv HTML 中可靠披露"
      ],
      "name": "Carlos Busso"
    }
  ],
  "contract": "researcher-sidecars-v1",
  "id": "2609.02940",
  "originalTitle": "Listen to the Latents: Self-Correcting Speech Recognition in Large Audio Language Models Through Hidden-State Interactions",
  "pdfUrl": "https://arxiv.org/pdf/2609.02940.pdf",
  "schemaVersion": 1,
  "type": "preprint"
}