Tag with "Text-to-Speech"

Google has unveiled Gemini TTS 2.5, a major upgrade to its text-to-speech technology that adds emotional expression and multi-language support. The new system lets voices shift tone instantly, adapt pacing to context, and maintain character consistency across 24 languages. Early adopters report significant improvements in user engagement while cutting costs. Available now for testing, the production version is expected in early 2025.

Microsoft has quietly released VibeVoice-Realtime-0.5B, a surprisingly nimble text-to-speech model that responds in just 300 milliseconds - faster than most humans can blink. This lightweight yet powerful tool can handle marathon 90-minute readings without missing a beat, juggle four distinct character voices simultaneously, and even detect emotions in text. While its English performance shines, the Chinese version still needs some polish. Already available on HuggingFace with an MIT license, developers are quickly integrating it into everything from audiobook apps to real-time translation tools.

The KittenML team has released Kitten TTS, an open-source text-to-speech model with just 15 million parameters and a size under 25MB. Designed for efficiency, it supports CPU-only operation, offers high-quality voice options, and promises real-time synthesis. Future releases include mobile SDKs and web versions.

AI entrepreneur Li Mu and Boson.ai have launched Higgs Audio v2, an advanced open-source text-to-speech model. Capable of multilingual dialogue, voice cloning, and automatic rhythm adjustment, it sets new benchmarks in speech synthesis. Trained on 10 million hours of data, it outperforms competitors in emotion and question categories. The model is now available on GitHub and Hugging Face.

Resemble AI has open-sourced Chatterbox, a powerful text-to-speech model that rivals ElevenLabs in performance. With advanced features like voice cloning and emotional control, it's set to transform voice technology applications while maintaining security through neural watermarking.

ElevenLabs has launched new features that enable users to create personalized conversational AI agents. Users can customize parameters such as voice tone and response length, integrate knowledge bases, and utilize different language models. The startup aims to enhance its competitive position in the AI space with these developments while planning to raise significant funding.

Oute AI has launched OuteTTS-0.1-350M, a groundbreaking text-to-speech synthesis method featuring zero-shot voice cloning capabilities. This approach simplifies TTS processes by eliminating complex architectures, making it efficient for real-time applications. The model is designed for accessibility and performance, appealing to various sectors such as personalized assistants and audiobooks.