Ken Sutton and Hamid Nawab, co-founders of Yobe, join us to discuss what the meaning of ‘cocktail party problem’ is and how to crack it in with voice AI.
Ever wondered why you in-car voice assistant never hears what you say? Ever caught yourself shouting into your headphones on a busy road, trying to wake up your assistant? One of the most challenging parts of voice AI is picking out voices in noisy environments.
You can retrain speech recognition models using acoustic modelling and invest in beamforming microphones, but that might not be enough. Today, we meet the company that has cracked voice recognition in extremely noisy environments.
Yobe co-founder and CEO, Ken Sutton, and Hamid Nawab, Co-founder and Chief Scientist, joins us to share more on how he and his team have cracked the ‘cocktail party problem’.
In this episode
00:00 Introduction and presenting VUX World @ VOICE22 Presented by Kore AI
Find out more at https://voicesummit.ai/agenda and save 20% on tickets with the code VUX20
03:29 The journey to Yobe
09:00 What is the cocktail party problem in AI?
19:18 Product market fit for inferential AI
25:15 What is it that Yobe does
30:00 How to process an audio signal without frequency manipulation
37:12 Using two microphones
42:35 The ideal clients
49:40 Intelligence in AI
Up your CX maturity in this free workshop with yours truly and Cognigy. Find out more: https://vux.world/cognigy
Find out more about VUX @ VOICE22 presented by Kore AI: www.voicesummit.ai and save 20% on tickets with the code VUX20