SAM-Translator

SAM-Translator is a proposed system designed to convert text from a source language directly into speech in a target language.

1. CVSS Dataset
2. CoVoST2 Dataset

This demo shows samples from the CVSS and CoVoST2 datasets. You can explore the translations by selecting different languages from the dropdown.

Speech Translation: Speech is synthesized by our proposed SAM-Translator, and the text is transcribed using a pretrained ASR (Whisper) model.
Speech Ground Truth: Speech and text are taken from the ground-truth dataset for reference.

CVSS Dataset

Select Language (CVSS):

Loading CVSS data...

CoVoST2 Dataset

Select Language (CoVoST2):

Loading CoVoST2 data...

SAM-Translator

Table of Contents

CVSS Dataset

CoVoST2 Dataset