Direct Simultaneous Speech-to-Speech Translation with Variational Monotonic Multihead Attention

TL;DR

This paper introduces a direct speech-to-speech translation model that uses discrete units and a novel variational monotonic multihead attention mechanism, improving translation quality and latency tradeoffs.

Contribution

The paper proposes a new direct Simul-S2ST model with V-MMA, enabling efficient policy learning and eliminating the need for intermediate text representations.

Findings

Direct model outperforms cascaded approach in quality-latency tradeoff.

Uses discrete units learned in an unsupervised manner for speech synthesis.

V-MMA improves efficiency in simultaneous translation policy learning.

Abstract

We present a direct simultaneous speech-to-speech translation (Simul-S2ST) model, Furthermore, the generation of translation is independent from intermediate text representations. Our approach leverages recent progress on direct speech-to-speech translation with discrete units, in which a sequence of discrete representations, instead of continuous spectrogram features, learned in an unsupervised manner, are predicted from the model and passed directly to a vocoder for speech synthesis on-the-fly. We also introduce the variational monotonic multihead attention (V-MMA), to handle the challenge of inefficient policy learning in speech simultaneous translation. The simultaneous policy then operates on source speech features and target discrete units. We carry out empirical studies to compare cascaded and direct approach on the Fisher Spanish-English and MuST-C English-Spanish datasets. Direct simultaneous model is shown to outperform the cascaded model by achieving a better tradeoff between translation quality and latency.