Gradium

Gradium is a Paris-based AI voice startup that builds ultra-low latency voice AI models for developers. Its products are delivered as real-time, scalable voice APIs: an expressive text-to-speech API designed for ultra-realistic, emotionally expressive speech; an accurate speech-to-text API; and high-fidelity voice cloning. The company also offers a voice agent platform intended to power voice agents across multiple industries.

Underpinning these products is a family of audio-native foundational language models. These unify generation, transcription, transformation and dialogue within a single neural architecture, designed to combine expressive, realistic speech with low latency while remaining efficient and scalable. The company's technical work spans generative audio, neural audio codecs, audio language models, low-latency inference and dialogue systems.

The founding team comprises researchers who collectively invented and published the methods and algorithms behind many of the voice and audio models in use today, with published work in neural audio codecs and audio language models. Gradium reports that it began serving customers in production within months of launch. Its stated mission is to make high-quality voice broadly affordable and widely available, and its platform is used in sectors including health, customer support, market research, gaming and digital advertisement.