Alex Smola, machine learning researcher and co-founder of Boson AI in 2023, is preparing to launch Higgs RealTime, a real-time speech-to-speech model. The Santa Clara-based startup released Higgs TTS 3 on June 4, 2026, supporting over 100 languages with zero-shot voice cloning and inline emotion control capabilities. Boson AI open-sourced its Higgs TTS 2 models on Hugging Face in May 2025 after training them on more than 10 million hours of audio data. Higgs RealTime eliminates the intermediate text conversion step, reducing latency while preserving vocal nuance. Competing against OpenAI, Google and ElevenLabs, Boson AI differentiates through academic expertise and open-source adoption strategy.
Source: Read the original article
