Gradium is a Paris-based AI voice startup that builds ultra-low latency voice AI models for developers. The company's core offering centres on foundational audio language models - a unified neural architecture designed for generation, transcription, transformation, and dialogue. These models aim to deliver emotionally expressive speech with minimal latency, packaged as real-time, scalable voice APIs.
The company's technical work spans generative audio, voice cloning, neural audio codecs, and audio language modelling. Its APIs provide expressive text-to-speech, accurate speech-to-text, and high-fidelity voice cloning capabilities, targeting developers integrating voice features into applications.
Gradium serves customers across several verticals, including health, customer support, market research, gaming NPCs, and digital advertisement. The company reports that it has moved to serving customers in production within months of its launch.
