Speechmatics, a deep tech firm, has partnered with NVIDIA to upgrade its machine learning capabilities using the company's accelerated computing systems and software stack. The partnership enables Speechmatics to deploy larger models, increasing its speech-to-text API's processing power and efficiency. With this upgrade, Speechmatics aims to improve its self-supervised learning technology, which requires huge datasets for training and large amounts of computation. The use of NVIDIA's DGX H100 system with eight new Tensor Core graphics processing units provides a step change in the speed and efficiency of training neural networks. This partnership is critical as speech-to-text becomes increasingly integral to various industries, including contact centers, media, and event captioning, among others.