Fake it to make it: Using synthetic data to remedy the data shortage in joint multimodal speech-and-gesture synthesis

TL;DR

This paper introduces a synthetic data generation approach using unimodal models to enhance joint speech and gesture synthesis from text, addressing data scarcity and improving output quality with a new controllable prosody architecture.

Contribution

It presents a novel method of synthesizing large-scale multimodal training data from unimodal models and proposes an improved synthesis architecture with better prosody control.

Findings

Pre-training on synthetic data enhances speech and gesture quality.

The new architecture improves controllability and synthesis performance.

Synthetic data pre-training yields significant benefits over limited real data.

Abstract

Although humans engaged in face-to-face conversation simultaneously communicate both verbally and non-verbally, methods for joint and unified synthesis of speech audio and co-speech 3D gesture motion from text are a new and emerging field. These technologies hold great promise for more human-like, efficient, expressive, and robust synthetic communication, but are currently held back by the lack of suitably large datasets, as existing methods are trained on parallel data from all constituent modalities. Inspired by student-teacher methods, we propose a straightforward solution to the data shortage, by simply synthesising additional training material. Specifically, we use unimodal synthesis models trained on large datasets to create multimodal (but synthetic) parallel training data, and then pre-train a joint synthesis model on that material. In addition, we propose a new synthesis architecture that adds better and more controllable prosody modelling to the state-of-the-art method in the field. Our results confirm that pre-training on large amounts of synthetic data improves the quality of both the speech and the motion synthesised by the multimodal model, with the proposed architecture yielding further benefits when pre-trained on the synthetic data. See https://shivammehta25.github.io/MAGI/ for example output.