loudness-tts ranks open Piper voices and normalizes text for Python and the browser
Open text-to-speech models now run quickly on ordinary hardware, but their output often sounds wrong. The developers of Loudness.ai built loudness-tts to address voice quality, licensing, and text handling problems. They tested 1,683 Piper voices for naturalness and clarity, and the voice auto-selection picks the best voice with a commercially usable licence. A rule layer converts money, dates, units, and abbreviations into spoken words in English and Spanish, while keeping word timings aligned with the source text for subtitles. The project offers matching Python and JavaScript editions, the latter running in Node, Bun, Deno, and the browser, and it is free and open source.
This is an AI-generated summary. ShortSingh links to the original source for the complete article.

Discussion (0)
Log in to join the discussion and vote.
Log in