MARS6: A Small and Robust Hierarchical-Codec Text-to-Speech Model

TL;DR

MARS6 is a compact, hierarchical transformer-based TTS model that achieves high-quality, expressive speech synthesis and zero-shot voice cloning with improved efficiency and stability, comparable to larger models.

Contribution

The paper introduces MARS6, a small hierarchical-encoder decoder transformer for TTS that enhances expressiveness, stability, and efficiency using recent modeling and training techniques.

Findings

Achieves high-quality TTS comparable to larger models.

Improves stability and reduces repetitive outputs.

Efficient processing of long-form text at 12 Hz.

Abstract

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/