Kensho Collaborates With NVIDIA to Advance Automatic Speech Recognition
Kensho, an S&P Global company, introduced Scribe, an end-to-end speech recognition solution specifically optimized for the finance and business community. Scribe leverages the latest deep learning techniques to process more audio in less time with better accuracy.
Over the past decade, S&P Global Market Intelligence, a division of S&P Global, has transcribed more than 100,000 hours of earnings calls. Using NVIDIA NeMo on NVIDIA GPUs, Kensho trained Scribe to systematically transcribe tens of thousands of earnings calls, management presentations, and acquisition calls each year – improving accuracy and enabling S&P Global to increase earnings call coverage by more than 25%.
Recommended AI News: Kahoot! Launches The New Interactive Presentations Feature To Make Any Presentation Awesome
“Technology and data science are playing a crucial role in how we bring value to our customers,” said Martina Cheung, President of S&P Global Market Intelligence. “Kensho’s collaboration with NVIDIA exemplifies S&P Global’s commitment to leveraging state-of-the-art machine learning to deliver our clients the essential intelligence they need to make informed business decisions with conviction.”
Recommended AI News: NVIDIA Jarvis Simplifies Building State-of-the-Art Conversational AI Services
Scribe is purpose-built to handle the complexities of financial audio – including financial jargon, product names and industry-specific terminology. Scribe’s syntactic capabilities seamlessly enable enumeration, capitalization and determination of sentence structure, while significantly outperforming other commercial solutions on business and financial audio. Scribe, among a suite of other AI-enabled capabilities designed to streamline workflows, will soon be available on the S&P Global Marketplace.
Kensho and NVIDIA are actively exploring further opportunities for collaboration. Kensho is already experimenting with NVIDIA NeMo acoustic model, which helps researchers quickly integrate model components, pushing the envelope of automatic speech recognition even further.
Recommended AI News: Megaport Deploys New SDN Cloud On-Ramp in EdgeConneX Warsaw Data Center Facility
Comments are closed, but trackbacks and pingbacks are open.