Compact Neural TTS Voices for Accessibility

TL;DR

This paper presents a compact neural TTS system that achieves high-quality speech synthesis with very low latency and small disk footprint, suitable for deployment on low-power devices for accessibility.

Contribution

A novel compact neural TTS model that balances naturalness, latency, and size, enabling real-time speech synthesis on resource-constrained devices.

Findings

Achieves approximately 15 ms latency.

Maintains high speech naturalness.

Has a low disk footprint suitable for handheld devices.

Abstract

Contemporary text-to-speech solutions for accessibility applications can typically be classified into two categories: (i) device-based statistical parametric speech synthesis (SPSS) or unit selection (USEL) and (ii) cloud-based neural TTS. SPSS and USEL offer low latency and low disk footprint at the expense of naturalness and audio quality. Cloud-based neural TTS systems provide significantly better audio quality and naturalness but regress in terms of latency and responsiveness, rendering these impractical for real-world applications. More recently, neural TTS models were made deployable to run on handheld devices. Nevertheless, latency remains higher than SPSS and USEL, while disk footprint prohibits pre-installation for multiple voices at once. In this work, we describe a high-quality compact neural TTS system achieving latency on the order of 15 ms with low disk footprint. The proposed solution is capable of running on low-power devices.