Saved in:
Bibliographic Details
Main Authors: Kumar, Sahil, Patel, Namrataben, Wang, Honggang, Zhang, Youshan
Format: Preprint
Published: 2026
Subjects:
Online Access:https://arxiv.org/abs/2604.00292
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866918421547974656
author Kumar, Sahil
Patel, Namrataben
Wang, Honggang
Zhang, Youshan
author_facet Kumar, Sahil
Patel, Namrataben
Wang, Honggang
Zhang, Youshan
contents MambaVoiceCloning (MVC) asks whether the conditioning path of diffusion-based TTS can be made fully SSM-only at inference, removing all attention and explicit RNN-style recurrence layers across text, rhythm, and prosody, while preserving or improving quality under controlled conditions. MVC combines a gated bidirectional Mamba text encoder, a Temporal Bi-Mamba supervised by a lightweight alignment teacher discarded after training, and an Expressive Mamba with AdaLN modulation, yielding linear-time O(T) conditioning with bounded activation memory and practical finite look-ahead streaming. Unlike prior Mamba-TTS systems that remain hybrid at inference, MVC removes attention-based duration and style modules under a fixed StyleTTS2 mel-diffusion-vocoder backbone. Trained on LJSpeech/LibriTTS and evaluated on VCTK, CSS10 (ES/DE/FR), and long-form Gutenberg passages, MVC achieves modest but statistically reliable gains over StyleTTS2, VITS, and Mamba-attention hybrids in MOS/CMOS, F0 RMSE, MCD, and WER, while reducing encoder parameters to 21M and improving throughput by 1.6x. Diffusion remains the dominant latency source, but SSM-only conditioning improves memory footprint, stability, and deployability.
format Preprint
id arxiv_https___arxiv_org_abs_2604_00292
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle MambaVoiceCloning: Efficient and Expressive Text-to-Speech via State-Space Modeling and Diffusion Control
Kumar, Sahil
Patel, Namrataben
Wang, Honggang
Zhang, Youshan
Sound
Machine Learning
MambaVoiceCloning (MVC) asks whether the conditioning path of diffusion-based TTS can be made fully SSM-only at inference, removing all attention and explicit RNN-style recurrence layers across text, rhythm, and prosody, while preserving or improving quality under controlled conditions. MVC combines a gated bidirectional Mamba text encoder, a Temporal Bi-Mamba supervised by a lightweight alignment teacher discarded after training, and an Expressive Mamba with AdaLN modulation, yielding linear-time O(T) conditioning with bounded activation memory and practical finite look-ahead streaming. Unlike prior Mamba-TTS systems that remain hybrid at inference, MVC removes attention-based duration and style modules under a fixed StyleTTS2 mel-diffusion-vocoder backbone. Trained on LJSpeech/LibriTTS and evaluated on VCTK, CSS10 (ES/DE/FR), and long-form Gutenberg passages, MVC achieves modest but statistically reliable gains over StyleTTS2, VITS, and Mamba-attention hybrids in MOS/CMOS, F0 RMSE, MCD, and WER, while reducing encoder parameters to 21M and improving throughput by 1.6x. Diffusion remains the dominant latency source, but SSM-only conditioning improves memory footprint, stability, and deployability.
title MambaVoiceCloning: Efficient and Expressive Text-to-Speech via State-Space Modeling and Diffusion Control
topic Sound
Machine Learning
url https://arxiv.org/abs/2604.00292