Microsoft VibeVoice TTS LOCAL Testing – A Multi-Speaker Podcast TTS!
Автор: Bijan Bowen
Загружено: 2025-08-25
Просмотров: 16622
Timestamps:
00:00 - Intro
01:35 - Technical Look & Local Setup
04:30 - First Test
04:58 - Multi Speaker Testing
07:50 - Four Speaker Test
08:46 - Weird Result
09:43 - Singing Test
11:50 - Disturbing Result
13:00 - 7B Testing
14:41 - Unintentional Singing
15:17 - 7B Singing Test
17:34 - 1.5B vs 7B Podcast Test
20:40 - Closing Thoughts
AI Integration & Consulting: https://bijanbowen.com
Join the Discord: / discord
In this video, we take a first look at the newly released VibeVoice TTS model family from Microsoft. This set of text-to-speech models is designed for extended, high-quality generation — and particularly for multi-speaker, podcast-style dialogue synthesis.
We start with a technical overview of the models and cover local setup on a test system. From there, we test both the 1.5B and 7B variants in a variety of use cases including multi-speaker dialogues, podcast simulations, and even singing — with some interesting and unexpected results along the way.
HF Link: https://huggingface.co/microsoft/Vibe...
Доступные форматы для скачивания:
Скачать видео mp4
-
Информация по загрузке: