3 Matching Annotations
  1. Last 7 days
  2. arxiv.org arxiv.org
    Improving Robustness of LLM-based Speech Synthesis by Learning Monotonic Alignment
    2
    1. bharat3012 30 Jan 2026
      n encoder-decodertransformers, the TTS alignment is learned in certain cross-attention heads of the decoder; while in decoder-only models,the alignment is learned in the self-attention layers.

      Good point of difference between En-De vs De only models.

    Visit annotations in context

    Annotators

    URL

    arxiv.org/pdf/2406.17957
  3. Jun 2024
  4. nvidia.github.io nvidia.github.io
    Using the NVIDIA API Catalog — NVIDIA Generative AI Examples 24.6.0 documentation
    1
    Visit annotations in context

    Annotators

    URL

    nvidia.github.io/GenerativeAIExamples/latest/api-catalog.html