Reproduce “Holistic Parallel Supervision for Expressive Speech to Speech Translation” on OpenResearch

Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.

Universal app for Apple silicon and Intel. Requires macOS 11 or later.

About this paper

Hao Zhang, Ismail Rasim Ulgen, Bismarck Bamfo Odoom, Rui Liu, Zihan Zhang, Berrak Sisman, Philipp Koehn

The authors introduce a speech-to-speech translation method that uses professional dubbing and voice conversion to create paired training data: converted dubbed speech is the source, while the original performance remains the target. In a small English-to-Spanish listening test, DirectS2ST trained on this data received an expressiveness rating of 3.42, versus 2.13 with CVSS-T data, supporting a benefit on those samples.

Summary by alphaXiv.

Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.