We present a noisy channel generative model of two sequences, for exampl...
We introduce the Globally Normalized Autoregressive Transducer (GNAT) fo...
This work explores the task of synthesizing speech in nonexistent
human-...
Non-saturating generative adversarial network (GAN) training is widely u...
In this technical report we describe some properties of f-divergences an...
Despite the ability to produce human-level speech for in-domain text,
at...
We present a novel generative model that combines state-of-the-art neura...
Recent work has explored sequence-to-sequence latent variable models for...
State-level minimum Bayes risk (sMBR) training has become the de facto
s...