Deep Voice 3: Scaling Text-to-Speech with Convolutional Sequence Learning

TLDR

Deep Voice 3 is a fully‑convolutional attention‑based neural text‑to‑speech system. Deep Voice 3 is trained on over 800 hours of audio from more than 2,000 speakers, incorporates techniques to mitigate attention‑based synthesis errors, compares multiple waveform synthesis methods, and scales inference to ten million queries per day on a single GPU. Deep Voice 3 matches state‑of‑the‑art neural TTS in naturalness while training ten times faster, and it identifies and mitigates common attention‑based synthesis errors while evaluating waveform synthesis approaches.

Abstract

We present Deep Voice 3, a fully-convolutional attention-based neural text-to-speech (TTS) system. Deep Voice 3 matches state-of-the-art neural speech synthesis systems in naturalness while training ten times faster. We scale Deep Voice 3 to data set sizes unprecedented for TTS, training on more than eight hundred hours of audio from over two thousand speakers. In addition, we identify common error modes of attention-based speech synthesis networks, demonstrate how to mitigate them, and compare several different waveform synthesis methods. We also describe how to scale inference to ten million queries per day on one single-GPU server.