{
  "absUrl": "https://arxiv.org/abs/2609.16514v1",
  "arxivId": "2609.16514",
  "arxivVersion": 1,
  "arxivVersionedId": "2609.16514v1",
  "authors": [
    {
      "affiliations": [
        "机构信息未在 arXiv HTML 中可靠披露"
      ],
      "name": "Qian Chen"
    },
    {
      "affiliations": [
        "机构信息未在 arXiv HTML 中可靠披露"
      ],
      "name": "Xiangang Li"
    },
    {
      "affiliations": [
        "机构信息未在 arXiv HTML 中可靠披露"
      ],
      "name": "Xiang Lv"
    },
    {
      "affiliations": [
        "机构信息未在 arXiv HTML 中可靠披露"
      ],
      "name": "Han Zhao"
    },
    {
      "affiliations": [
        "Alibaba Token Foundry",
        "Equal contribution; alphabetical by last name."
      ],
      "name": "Tianyu Zhao"
    }
  ],
  "contract": "researcher-sidecars-v1",
  "id": "2609.16514v1",
  "originalTitle": "The Evolving Bottleneck in Speech Generation: Interface Co-design and Staged Alignment from CosyVoice to Qwen-Audio-3.0-TTS",
  "pdfUrl": "https://arxiv.org/pdf/2609.16514v1.pdf",
  "schemaVersion": 1,
  "type": "preprint"
}