Export

LJSpeech format: a zip with a wavs/ directory plus a pipe-delimited metadata.csv of clip_id|transcript, ordered by emotional register then duration.