Saved in:
Bibliographic Details
Main Authors: Baas, Matthew, Scholtz, Pieter, Mehta, Arnav, Dyson, Elliott, Prakash, Akshat, Kamper, Herman
Format: Preprint
Published: 2025
Subjects:
Online Access:https://arxiv.org/abs/2501.05787
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866915097144721408
author Baas, Matthew
Scholtz, Pieter
Mehta, Arnav
Dyson, Elliott
Prakash, Akshat
Kamper, Herman
author_facet Baas, Matthew
Scholtz, Pieter
Mehta, Arnav
Dyson, Elliott
Prakash, Akshat
Kamper, Herman
contents Codec-based text-to-speech (TTS) models have shown impressive quality with zero-shot voice cloning abilities. However, they often struggle with more expressive references or complex text inputs. We present MARS6, a robust encoder-decoder transformer for rapid, expressive TTS. MARS6 is built on recent improvements in spoken language modelling. Utilizing a hierarchical setup for its decoder, new speech tokens are processed at a rate of only 12 Hz, enabling efficient modelling of long-form text while retaining reconstruction quality. We combine several recent training and inference techniques to reduce repetitive generation and improve output stability and quality. This enables the 70M-parameter MARS6 to achieve similar performance to models many times larger. We show this in objective and subjective evaluations, comparing TTS output quality and reference speaker cloning ability. Project page: https://camb-ai.github.io/mars6-turbo/
format Preprint
id arxiv_https___arxiv_org_abs_2501_05787
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle MARS6: A Small and Robust Hierarchical-Codec Text-to-Speech Model
Baas, Matthew
Scholtz, Pieter
Mehta, Arnav
Dyson, Elliott
Prakash, Akshat
Kamper, Herman
Audio and Speech Processing
Computation and Language
Codec-based text-to-speech (TTS) models have shown impressive quality with zero-shot voice cloning abilities. However, they often struggle with more expressive references or complex text inputs. We present MARS6, a robust encoder-decoder transformer for rapid, expressive TTS. MARS6 is built on recent improvements in spoken language modelling. Utilizing a hierarchical setup for its decoder, new speech tokens are processed at a rate of only 12 Hz, enabling efficient modelling of long-form text while retaining reconstruction quality. We combine several recent training and inference techniques to reduce repetitive generation and improve output stability and quality. This enables the 70M-parameter MARS6 to achieve similar performance to models many times larger. We show this in objective and subjective evaluations, comparing TTS output quality and reference speaker cloning ability. Project page: https://camb-ai.github.io/mars6-turbo/
title MARS6: A Small and Robust Hierarchical-Codec Text-to-Speech Model
topic Audio and Speech Processing
Computation and Language
url https://arxiv.org/abs/2501.05787