VieNeu-TTS

Why we included this project

VieNeu-TTS runs entirely on your own hardware, doing real-time Vietnamese speech synthesis on CPU, so you never touch per-character cloud pricing. Beyond plain text-to-speech, it can clone a voice from a three to eight second clip, switch between Vietnamese and English mid-sentence, and render a multi-speaker podcast or conversation script as a single batch. The newer v2 model ships with default voices, which means you don't need a reference recording to get started, and offers reading styles like news and storytelling plus experimental emotion tags you can embed in the text. That combination suits interactive voice products, offline demo kiosks, and localization pipelines where latency and data privacy rule out the cloud. A Web UI and a straightforward Python SDK, covering v2 and the early-access v3 Turbo backbone, let teams try it out quickly before committing to production.