SoundHound Partners with NVIDIA to Bring Natural Language Understanding to Cars

SoundHound, a provider of sound recognition and voice search technologies, is partnering with NVIDIA to create a large-vocabulary, hybrid voice and natural language understanding interface for in-vehicle infotainment systems that is based on deep learning.

The solution uses the Houndify speech recognition engine, the NVIDIA DRIVE platform, and SoundHound's patented technologies for running dual mode speech recognition simultaneously in the cloud and in embedded systems.

With the Houndify integration, drivers can interact with more than 100 applications that provide users with relevant information, such as directions, weather, stock prices, sports scores, flight status, local business searches, and hotel searches. They can also control in-car systems, such as heating/cooling, windows, music, or a phone.

"Our mission is Houndify everything, to add a smart, voice-enabled, conversational interface to every technology that humans interact with," said Keyvan Mohajer, founder and CEO of SoundHound, in a statement. "Enabling drivers to interact with their cars using voice and natural language is extremely compelling and increases safety and productivity. Key to this ability are NVIDIA GPUs used for deep learning training, and the DRIVE CX platform, with its powerful GPU for embedded computation. We are able to run Houndify's large vocabulary speech recognition and natural language understanding in the car on DRIVE CX, even without a cloud connection."

SoundHound recently launched the Houndify developer platform. Some partner companies, including Expedia, AccuWeather, Sportradar, Xignite, FlightStats, and Open Exchange Rates, are providing their own data and APIs to the Houndify platform so developers can integrate with their services. These domains are already integrated within the DRIVE CX platform. The first application to showcase the Houndify platform is SoundHound's own voice search and assistant app, Hound, which launched in private beta in June.

NVIDIA provides end-to-end deep learning solutions for both the cloud and in-vehicle applications. A single architecture enables deep neural networks to be trained in the datacenter and then deployed in the car for various applications from autonomous driving to natural language processing. The company has provided in-vehicle infotainment and digital cockpit technology to more than 10 million cars on the road today.

"Using NVIDIA's GPUs and deep learning technology, Houndify has demonstrated great progress in enabling fast speech recognition technology, even when disconnected from the Internet," said Rob Csongor, vice president and general manager of automotive at NVIDIA, in a statement. "Using deep learning, the system is able to rapidly understand complex natural language queries while supporting a large and growing number of search categories."

SpeechTek Covers
for qualified subscribers
Subscribe Now Current Issue Past Issues
Related Articles

SoundHound Adds Uber and Yelp as Houndify Partners

Data form Uber and Yelp will be available from Uber and Yelp to developers using the Houndify voice platform.