A collection of high-quality speech datasets with transcripts in multiple languages.