Case Study: Krisp speeds up real-time voice AI training with Nebius AI

A Nebius Case Study

Preview of the Krisp Case Study

Krisp’s advancements in real‑time voice AI

Krisp, a leading AI-powered noise cancellation technology company, needed a powerful and reliable cloud platform to train the large foundational models for its real-time Accent Localization technology. They faced the challenge of needing to run numerous training experiments on high-performance GPUs to develop an optimal model that could later be distilled into a highly efficient version for low-powered call center computers.

Nebius AI provided Krisp with its high-performance, AI-centric cloud solution, specifically utilizing NVIDIA H100 GPUs. This infrastructure allowed Krisp to train its foundational model up to 80% faster than on previous hardware, enabling the team to meet a tight product development timeline. The Nebius platform was noted for its stability and excellent support, with Krisp able to begin transferring data and training within a day of the initial discussion. The collaboration provided valuable insights for optimizing training processes and successfully advanced the development of Krisp's on-device AI solutions.


View this case study…

Nebius

14 Case Studies