Speechmatics’ new high-performance computing (HPC) deployment includes NVIDIA A100 GPUs, housed within an advanced Supermicro computing infrastructure, which will be hosted at Kao Data’s Harlow campus. The supercomputer will allow Speechmatics to expand its GPU-accelerated neural network research and development, supporting increasing customer demand for its leading speech recognition technology.
Kao Data was chosen as the UK’s premier location for HPC and AI, in addition to its commitments to sustainability and energy efficiency. Speechmatics’ high-density supercomputer will benefit from bespoke colocation, an SLA-backed PUE of 1.2, and will be powered by 100 percent renewable energy.
Recognised in both 2020 and 2021 by the FT1000 as one Europe’s fastest-growing companies, and named within NVIDIA’s prestigious Inception Program, Speechmatics exploits deep learning technology to deliver highly accurate and ultra-fast speech-to-text technology. The company utilises NVIDIA’s acclaimed GPU hardware to complete its aim of understanding every voice, regardless of demographic, accent, or dialect.
Speechmatics’ supercomputer will work in tandem with its hyperscale cloud deployments, providing a pioneering example of fluent, hybrid HPC in-action. Central to this capability is Megaport’s hyperscale connectivity solutions at Kao Data’s Harlow campus, which provides seamless connectivity and on-ramps between the on-premises supercomputer, and Speechmatics’ instances within Amazon Web Services (AWS), Microsoft Azure, and Google Cloud.
“Powering modern Machine Learning research and development requires an advanced computing infrastructure which only becomes more demanding as you scale.” Said Will Williams, VP Machine Learning, Speechmatics. “As we continue to develop our technological edge through the use of self-supervised learning in our models, it’s crucial to ensure our data centre provider can scale with our compute demands, but in a sustainable way. Kao Data was the obvious choice for us as the best HPC data centre operator in the UK with the benefit of being powered by 100% renewable energy.”
“The opportunity to support Speechmatics in this crucial phase of the company’s expansion is an exciting prospect for our organisation, and further underpins Kao Data’s position as the UK’s preeminent provider of sustainable data centres for HPC and AI,” said Lee Myall, CEO, Kao Data. “We look forward to working closely with Speechmatics to help power and support their unique approach of applying neural networks to speech recognition.”
Speechmatics’ founder, Dr Tony Robinson, pioneered the approach of applying neural networks to speech recognition back in the 1980s, and when higher performance CPUs appeared in the early 2000’s, the company was among the first developers to utilise the technology to process large datasets at volume. Today, NVIDIA GPUs have further accelerated Speechmatics’ speech-to-text capabilities, and its ability to support leading organisations such as Deloitte, Veritone and Red Bee Media. The company’s flexible API easily integrates into customers services, solutions, and applications to offer the most accurate AI-powered transcription capability, with 33 languages already available.