Word to Speech, With the Subtitles Already Lined Up

Field notes

Users enter plain UTF-8 text, select a voice, and receive an audio file with caption files generated from the same timing data. The interface exposes a 5,000-character counter and a chronogram showing word timing from 0 to 3712 milliseconds. Exports are offered as mono 24 kHz / 48 kbps MP3, UTF-8 SRT, and WebVTT; the page also states that 322 voices and 150 languages are available.