{"ID":6620555,"CreatedAt":"2026-07-15T01:01:48.440468303Z","UpdatedAt":"2026-07-15T03:28:55.185153975Z","DeletedAt":null,"paper_url":"https://arxiv.org/abs/2607.12496","arxiv_id":"2607.12496","title":"ZipL-Dialog: Memory-Efficient Long-Form Spoken Dialog Synthesis via Latent Flow Matching","abstract":"Zero-shot dialog TTS benefits from flow-matching, but minute-scale generation on dense mel-spectrograms causes severe memory bottlenecks, often forcing unnatural chunked synthesis. We propose ZipL-Dialog, which shifts conditional flow-matching into a 4x time-compressed (25 Hz) latent space. To preserve acoustic fidelity under compression, we employ a deterministic mel autoencoder with auxiliary mel-domain supervision and optimize the ZipFormer's hierarchical downsampling schedule. Experiments show that ZipL-Dialog reduces maximum peak GPU memory by 11.22x and accelerates inference by 2.23x over the baseline, substantially lowering the memory footprint of single-pass multi-minute dialog synthesis while maintaining perceptual naturalness.","short_abstract":"Zero-shot dialog TTS benefits from flow-matching, but minute-scale generation on dense mel-spectrograms causes severe memory bottlenecks, often forcing unnatural chunked synthesis. We propose ZipL-Dialog, which shifts conditional flow-matching into a 4x time-compressed (25 Hz) latent space. To preserve acoustic fidelit...","url_abs":"https://arxiv.org/abs/2607.12496","url_pdf":"https://arxiv.org/pdf/2607.12496v1","authors":"[\"Jihwan Kim\",\"Nam Soo Kim\"]","published":"2026-07-14T08:28:48Z","proceeding":"eess.AS","tasks":"[\"eess.AS\"]","methods":"[]","has_code":false}