Issue link: https://iconnect007.uberflip.com/i/1181966
56 SMT007 MAGAZINE I NOVEMBER 2019 is something that's always in the back of my mind. It requires a decent amount of resources to put together a good cleanroom and equipment. If you don't have the bandwidth to assemble those packages in your service offerings, then it could mean losing business. Your customers might go somewhere else. Johnson: Let's consider some of the Tier 1 OEMs, as you mentioned earlier. As we break away from the more traditional manufacturing approaches and techniques—like a direct con- nection with the chips, etc.—do you see dif- ferent companies choosing different paths? Of course, there's a risk of taking a company's customer base as they move into new technol- ogies and fragmenting their customers into dif- ferent specialties? Is that a concern for EMS suppliers as we look at new technologies? Khan: There is some risk, and the Tier 1 EMS companies pretty much have the technology in their NPI or prototype divisions. But there is a risk factor as well. How many people see the changes in the technology that are happening? How many are willing to take the CapEx and put it into use by being able to put all of these types of infrastructures in place? And if you are not going to address that, then it will be a risk for you, and customers might go to some other places that are able to offer these PCB micro- electronics. Johnson: You raise a valid point there, Zulki. Thank you for your time. Khan: Thank you. SMT007 Eleven years ago, Carnegie Mellon University alumni Anthony Gadient, Edward Lin, and Rob Rutenbar were hun- kered down in a garage, chowing pizza over late nights of coding. Eighteen months later, voice startup Voci emerged as a spinout from CMU. Voci, like that of many early AI researchers, became a reality as a startup because of breakthroughs in deep neu - ral networks paired with advances in GPU computing. "Our academic roots are based on this idea that you can do bet- ter by taking advantage of application-specific hardware, such as NVIDIA GPUs," said Gadient, Voci's chief strategy officer and co-founder. Voci's V-Blaze automated speech recognition offers real- time speech-to-text and audio analytics to analyze con - versations between customers and call center represen- tatives. The data can be used by customers to understand the sentiment and emotion of speak- ers. Companies can use Voci to track what customers are saying about competitive products and different features offered else - where. Voci is also addressing a prob- lem that plagues automated cus- tomer service systems: caller verification. Many of these systems ask callers a handful of verification questions and then ask those same questions again if live support is required or if the call gets transferred. Instead, Voci has developed an API for "voiceprints" that can identify people by voice, bypassing the maze of verification questions. "Biometrics for voice is a problem worth solving, if only for our collective sanity. It enables machine verification of call - ers in the background instead of those maddening repeated questions you can face when handed off from operator to operator in a call center," said Gadient. Voci uses a multi- tude of neural networks and techniques to offer its natural language processing services. The service is offered either on-premises or in the cloud and taps into NVIDIA V100 Ten- sor Core GPUs for inference. Developers at Voci trained their networks on more than 20,000 hours of audio from customers seeking results for their busi- nesses. "It took approximately one month to train the neural nets on a network of machines run- ning a combination of NVIDIA P100 and V100 GPUs," said Gadient. (Source: NVIDIA) Heard Mentality: AI Voice Startup Helps Hear Customer Pain Points