SoundHound’s Chat AI & NVIDIA DRIVE Collaboration – Insights from COO Mike Zagorsek

Automotive Industries interview with for Mike Zagorsek, COO of SoundHound AI

In the rapidly evolving landscape of automotive technology, the fusion of artificial intelligence and vehicular functionality is reshaping the driving experience. Automotive Industries had the privilege of interviewing Mike Zagorsek, the COO of SoundHound AI, a pioneering force in this transformative field. The conversation delved into SoundHound’s groundbreaking collaboration with NVIDIA DRIVE and its impact on in-vehicle voice assistance.

Zagorsek elucidated on how SoundHound’s partnership with NVIDIA DRIVE heralds a new era in in-vehicle voice technology. By leveraging NVIDIA DRIVE, SoundHound’s Chat AI platform integrates real-time and generative AI capabilities, enabling sophisticated responses to voice queries even in offline scenarios, thereby redefining the boundaries of in-vehicle connectivity.

The dialogue then transitioned to the advantages of deploying large language models (LLMs) entirely on the edge, a feat accomplished by SoundHound. Zagorsek highlighted the trifecta of benefits: enhanced privacy, flexibility, and reduced operating costs. By empowering automakers to deliver hands-free access to powerful LLM capabilities without relying on cloud connectivity, SoundHound’s technology ensures data privacy, operational flexibility, and cost efficiency.

Further exploration illuminated SoundHound’s unique approach to in-vehicle voice platforms. Zagorsek elaborated on how SoundHound Chat AI seamlessly integrates real-time and generative AI capabilities, offering users a truly conversational experience. With the ability to intelligently select responses from diverse domains, SoundHound’s platform ensures accurate and contextual interactions, transcending the limitations of traditional voice assistants.

The conversation expanded to encompass SoundHound’s Vehicle Intelligence feature, which enhances the driving experience by providing comprehensive support beyond vehicle-related inquiries. Leveraging AI and LLMs, SoundHound’s Voice Intelligence domain delivers real-time, hands-free access to a myriad of vehicle-related information, from maintenance to troubleshooting, elevating user convenience and safety.

Zagorsek underscored SoundHound’s pioneering role in the conversational intelligence space, particularly within the automotive sector. With a rich legacy of innovation spanning two decades, SoundHound boasts unparalleled AI expertise and a robust technology stack. Their proprietary technologies, including Speech-to-Meaning and Deep Meaning Understanding, coupled with a commitment to data privacy and user experience, position SoundHound as a frontrunner in revolutionizing in-vehicle voice assistance.

In the Automotive Industries interview which follows Mike Zagorsek emphasized SoundHound’s global footprint, with Chat AI deployed in vehicles across 18 countries and 13 languages, cementing its status as a driving force in the automotive AI landscape.

Automotive Industries interview questions for Mike Zagorsek, COO of SoundHound AI

Automotive Industries: Hi Mike, can you explain how SoundHound’s partnership with NVIDIA DRIVE will revolutionize in-vehicle voice assistance?

Zagorsek:  NVIDIA DRIVE will enable SoundHound Chat AI – our breakthrough in-vehicle voice platform that combines both real-time and generative AI capabilities – to provide sophisticated responses to voice queries even when there is no access to the cloud.

Automotive Industries: What are the primary benefits of running a large language model completely on the edge, as opposed to relying on cloud connectivity?

Zagorsek:  This technology will allow automakers to give drivers and passengers handsfree access to powerful LLM capabilities with the added benefit of greater privacy (information doesn’t have to be sent to the cloud), flexibility (the ability to provide desired functionality with or without connectivity), and lower operating costs (lower cloud costs, which adds up over time).

Automotive Industries: Could you elaborate on how SoundHound’s in-vehicle voice platform combines both real-time and generative AI capabilities?

Zagorsek:  For a long time, SoundHound’s in-vehicle voice platform has provided access to hundreds of real-time information domains – from flight times, to weather, sports scores, navigation, local search, and many more. With SoundHound Chat AI, we built the very first platform that also grants voice-enabled access to generative AI via ChatGPT or almost any other available LLM.

In practice, this integration allows users to have a truly conversational experience because SoundHound Chat AI keeps the conversation going, providing fast, accurate responses to queries without frustrating results (no more “Sorry, I didn’t get that”). It does this using proprietary technology that intelligently selects the correct response from the most appropriate domain – whether that’s a ChatGPT-powered answer, or the kind of response that large language models can’t handle, like real-time questions about weather, sports, stock, and flight status.

Automotive Industries: How does SoundHound’s Vehicle Intelligence feature enhance the driving experience for users?

Zagorsek:  As technology develops, car manufacturers introduce more features every year – making car manuals increasingly complex. But in a fast-paced, digitized world drivers want instant access to information about their vehicle without having to pull over or spend time leafing through a physical manual. SoundHound’s voice-enabled Vehicle Intelligence domain uses artificial intelligence and large language models (LLMs) to provide answers to almost any question about a car’s functionality, maintenance, or repair – completely hands-free.

Automotive Industries: In what ways does the new in-vehicle voice assistant provide support for drivers beyond just vehicle-related inquiries?

Zagorsek:  SoundHound’s Vehicle Intelligence domain can leverage information from the manufacturer’s car manual for any brand and allows drivers to use regular conversational language to ask questions and receive answers in real-time. Some examples include:

• Can I start the engine remotely with my keys before getting in the car?
• What does the “auto hold” button do? 
• How do I see how much further I can drive with my current gas amount?
• Is there a way to adjust the height of the trunk door when it’s open?
• I see a flashing light that looks like a car battery and I’m not sure what that means?

Vehicle Intelligence is able to deliver fast, accurate responses to almost any vehicle-related query, including maintenance, safety, settings and customization, special features, vehicle systems, and general troubleshooting.

Automotive Industries: What advantages does this new technology offer automakers in terms of privacy, flexibility, and operating costs?

Zagorsek:  [as above] This technology will allow automakers to give drivers and passengers handsfree access to powerful LLM capabilities with the added benefit of greater privacy (information doesn’t have to be sent to the cloud), flexibility (the ability to provide desired functionality with or without connectivity), and lower operating costs (lower cloud costs, which adds up over time).

Automotive Industries:  How does the collaboration between SoundHound and NVIDIA enhance the occupant experience in vehicles?

Zagorsek:  When automakers can run voice assistants with generative AI capabilities on the edge, they allow their drivers to access sophisticated capabilities like Vehicle Intelligence and trip planning, even when the vehicle doesn’t have – or is unable to access – connectivity. This means being able to provide a consistency of service in remote locations or those with limited or intermittent access to the cloud.

Moreover, where customers have privacy concerns, they can be assured that their voice queries remain “on device” where the assistant is powered by an embedded platform, like NVIDIA DRIVE.

Automotive Industries: What sets SoundHound apart from other players in the conversational intelligence space, particularly in the automotive sector?

Zagorsek: SoundHound was the first to offer an in-vehicle voice assistant with generative AI capabilities. We were also the first in the world to pilot and then deploy with a major brand. At present, SoundHound Chat AI is live in vehicles across 18 countries in 13 languages.

The reason we were able to move so quickly is because we have deep AI expertise, and nearly 20 years of experience building best-in-class voice technology. As a company, we own our full technology stack, and hold over 200 patents – including for our proprietary Speech-to-Meaning, Deep Meaning Understanding technologies, which allow our assistants to process and understand voice queries with exceptional speed and accuracy. Additionally, SoundHound Chat AI deploys a proprietary arbitration technique that uses software engineering and machine learning to help prevent the system responding with risky and misleading “AI hallucinations.”

We have a well-established product and a history with the industry. SoundHound is currently working with 20 auto brands, representing around 25% of all units.