{
  "absUrl": "https://arxiv.org/abs/2609.04242",
  "arxivId": "2609.04242",
  "arxivVersion": null,
  "arxivVersionedId": null,
  "authors": [
    {
      "affiliations": [
        "机构信息未在 arXiv HTML 中可靠披露"
      ],
      "name": "Ankan Deria"
    },
    {
      "affiliations": [
        "机构信息未在 arXiv HTML 中可靠披露"
      ],
      "name": "Hanoona Rasheed"
    },
    {
      "affiliations": [
        "机构信息未在 arXiv HTML 中可靠披露"
      ],
      "name": "Xilin He"
    },
    {
      "affiliations": [
        "机构信息未在 arXiv HTML 中可靠披露"
      ],
      "name": "Fahad Shahbaz Khan"
    },
    {
      "affiliations": [
        "机构信息未在 arXiv HTML 中可靠披露"
      ],
      "name": "Salman Khan"
    }
  ],
  "contract": "researcher-sidecars-v1",
  "id": "2609.04242",
  "originalTitle": "Training-Free Speech-Centric Omni Understanding with Frozen VLMs",
  "pdfUrl": "https://arxiv.org/pdf/2609.04242.pdf",
  "schemaVersion": 1,
  "type": "preprint"
}