Case Study: Cedat 85 achieves 2.4x faster speech-to-text model training with IBM Power Systems AC922

Turning speech into a digital asset with speech-to-text solutions based on powerful cognitive technology

Cedat 85, a speech-to-text technology company, needed to speed up the training of its neural network models to keep pace with the complexity of accents, dialects, and languages in speech recognition. Using IBM Power Systems AC922 servers, Cedat 85 aimed to reduce weeks- or months-long model training cycles and strengthen its competitive position.

IBM implemented Power Systems AC922 servers to accelerate AI workloads and help Cedat 85 train models much faster without sacrificing accuracy. The result was 2.4x faster model training, nearly 10% higher accuracy in one government pilot, and the ability for Cedat 85 to take on more projects, bring solutions to market sooner, and grow revenue.

Cedat 85

Enrico Giannotti

Managing Director