Parrotron: An End-to-End Speech-to-Speech Conversion Model and its Applications to Hearing-Impaired Speech and Speech Separation

TL;DR

Parrotron is an end-to-end speech-to-speech conversion model that normalizes diverse speech inputs into a consistent target voice, improving intelligibility for hearing-impaired speech and enabling speech separation.

Contribution

This work introduces a novel end-to-end neural model for speech normalization and conversion, applicable to hearing-impaired speech and speech separation tasks.

Findings

Successfully normalizes speech across accents and noise conditions

Improves intelligibility and naturalness of deaf speech

Can be adapted for speech separation tasks

Abstract

We describe Parrotron, an end-to-end-trained speech-to-speech conversion model that maps an input spectrogram directly to another spectrogram, without utilizing any intermediate discrete representation. The network is composed of an encoder, spectrogram and phoneme decoders, followed by a vocoder to synthesize a time-domain waveform. We demonstrate that this model can be trained to normalize speech from any speaker regardless of accent, prosody, and background noise, into the voice of a single canonical target speaker with a fixed accent and consistent articulation and prosody. We further show that this normalization model can be adapted to normalize highly atypical speech from a deaf speaker, resulting in significant improvements in intelligibility and naturalness, measured via a speech recognizer and listening tests. Finally, demonstrating the utility of this model on other speech tasks, we show that the same model architecture can be trained to perform a speech separation task