Gradium

Paris-based Gradium builds audio language models and real-time voice AI for natural, low-latency spoken interaction.

Its technology spans speech generation, transcription and the voice-driven agents that companies increasingly want for phone and in-app experiences. Founded in 2024, it is part of a fast-moving European cohort tackling voice as a first-class AI modality.

The focus on latency reflects how much real-time responsiveness matters for voice to feel genuinely conversational.