Andrew Gibiansky

San Francisco, CA •

8 posts •

RSS

DiffWave and WaveGrad: Overview (Part 1)

DiffWave and WaveGrad: Overview (Part 1)

DiffWave and WaveGrad propose a new neural vocoder model based on diffusion probabilistic processes, with several nice properties and a solid theoretical justification.

Full Story

WaveNet and Tacotron aren't TTS systems

WaveNet and Tacotron aren't TTS systems

Deep learning models for speech synthesis, such as Google's WaveNet and Tacotron, are not complete text-to-speech systems.

Full Story

Recent Posts

Streaming Audio Synthesis

The naive approach to streaming audio synthesis using deep neural networks is to break up the input into chunks and then run synthesis on each chunk. Unfortunately, this introduces wasted computation and discontinuities. In this blog post, I present a simple and robust alternative.

Brainstorming: Neural Transducers for Speech Synthesis

Neural transducers are commonly used for automatic speech recognition (ASR), often achieving state-of-the-art results for

PQMF: Sub-band Coding for Neural Vocoders (Part 2)

This is a continuation of Part 1 of this two-part series. In this post, I'll