SwanVoice: Expressive Long-Form Zero-Shot Speech Synthesis for Both Monologue and Dialogue

Hugging Face Daily Papers Papers
zero-shot text-to-speech dialogue monologue speech-synthesis expressive long-form

Summary

SwanVoice is a zero-shot text-to-speech model designed for expressive long-form monologue and dialogue synthesis, combining VAE, flow-matching DiT, and diffusion post-training to achieve higher richness and hierarchy scores than existing baselines.

Zero-shot text-to-speech (TTS) has improved substantially for single-speaker synthesis, yet expressive long-form multi-speaker dialogue remains difficult. A common workaround is to synthesize each turn with a monologue TTS model and stitch the outputs together. This adds inference cost and often breaks acoustic consistency, conversational coherence, and affective continuity across turns. Recent dialogue TTS systems have begun to address this setting, but they still struggle to keep expressive coherence, controllable speaker switching, and monologue quality at the same time. We present SwanData-Speech and SwanVoice. SwanData-Speech builds monologue and dialogue corpora from in-the-wild audio, using Swan Forced Aligner for pause-aware word-level alignment and RobustMegaTTS3 for pronunciation-hard cases. Built on these data, SwanVoice is a zero-shot TTS model for 1--4 speakers, combining a 25 Hz VAE, raw-text conditioning with pause-aware symbols and pinyin substitution, and a flow-matching DiT with speaker-turn conditioning. Training starts from monologue speech, moves through mixed and real dialogue data, and then uses DiffusionNFT post-training with phone-level and speaker-similarity rewards. On SwanBench-Speech, SwanVoice obtains higher richness and hierarchy scores than all evaluated open-source baselines in both monologue and dialogue settings, while content accuracy remains the main limitation. Audio demos are available at https://swanaigc.github.io//#swanvoice.
Original Article
View Cached Full Text

Cached at: 06/01/26, 07:18 AM

Paper page - SwanVoice: Expressive Long-Form Zero-Shot Speech Synthesis for Both Monologue and Dialogue

Source: https://huggingface.co/papers/2605.30993

Abstract

A zero-shot text-to-speech system called SwanVoice is presented that addresses expressive long-form multi-speaker dialogue synthesis by combining VAE, flow-matching DiT, and diffusion post-training techniques.

Zero-shot text-to-speech(TTS) has improved substantially for single-speaker synthesis, yet expressive long-form multi-speaker dialogue remains difficult. A common workaround is to synthesize each turn with amonologue TTSmodel and stitch the outputs together. This adds inference cost and often breaks acoustic consistency, conversational coherence, and affective continuity across turns. Recentdialogue TTSsystems have begun to address this setting, but they still struggle to keep expressive coherence, controllable speaker switching, and monologue quality at the same time. We present SwanData-Speech and SwanVoice. SwanData-Speech builds monologue and dialogue corpora from in-the-wild audio, using SwanForced Alignerfor pause-aware word-level alignment and RobustMegaTTS3 for pronunciation-hard cases. Built on these data, SwanVoice is a zero-shot TTS model for 1--4 speakers, combining a 25 HzVAE, raw-text conditioning withpause-aware symbolsandpinyin substitution, and aflow-matching DiTwithspeaker-turn conditioning. Training starts from monologue speech, moves through mixed and real dialogue data, and then uses DiffusionNFT post-training with phone-level and speaker-similarity rewards. On SwanBench-Speech, SwanVoice obtains higher richness and hierarchy scores than all evaluated open-source baselines in both monologue and dialogue settings, while content accuracy remains the main limitation. Audio demos are available at https://swanaigc.github.io//#swanvoice.

View arXiv pageView PDFProject pageAdd to collection

Get this paper in your agent:

hf papers read 2605\.30993

Don’t have the latest CLI?curl \-LsSf https://hf\.co/cli/install\.sh \| bash

Models citing this paper0

No model linking this paper

Cite arxiv.org/abs/2605.30993 in a model README.md to link it from this page.

Datasets citing this paper0

No dataset linking this paper

Cite arxiv.org/abs/2605.30993 in a dataset README.md to link it from this page.

Spaces citing this paper0

No Space linking this paper

Cite arxiv.org/abs/2605.30993 in a Space README.md to link it from this page.

Collections including this paper0

No Collection including this paper

Add this paper to acollectionto link it from this page.

Similar Articles