audio model

August 27 news, technology media marktechpost on August 25 published a blog post, reported that Microsoft released open source text-to-speech (TTS) model VibeVoice-1.5B, which can generate up to 90 minutes at a time, up to four different speakers of the natural voice, and support cross-language and song synthesis. In terms of architecture, VibeVoice-1.5B is based on the Qwen2.5 language model with 1.5B parameters, combining acoustic (Acoustic) and semantic (Semantic) bi-partitioning ...

May 15 - TechCrunch published a blog post yesterday (May 14) reporting that AI startup Stability AI has launched Stable Audio Open Small, which it claims is the fastest "stereo" audio-generating AI model on the market, and can be run on smartphones. smartphones. Compared to other AI audio apps like Suno and Udio, Stable Audio Open Small doesn't need to rely on cloud processing, and its efficient design...