SoundHound Adds Generative AI Access to Voice Assistant Platform
SoundHound is joining the rush toward generative AI with an upgrade to the Dynamic Interaction conversational AI feature introduced in November. Dynamic Interaction’s voice AI for business services will now count generative AI in the style of ChatGPT as an option for the multi-modal interface.
SoundHound envisions generative AI as a way of enhancing the voice assistants it provides its clients. As an example in the demo video above, the car voice assistant powered by SoundHound doesn’t just give directions or a list of trails when the driver asks, “Can you help me plan a fun day of hiking in San Francisco for tomorrow?” The AI responds with an itinerary of trails that fit in a single route, describes a highlight from each trail, and suggests timing the last hike to end at sunset for the best view. That’s on top of Dynamic Interaction’s improved natural language understanding for fragmentary phrases and complementary visuals. The initial version also allows users to interrupt the AI to change or clarify something they said, a helpful trait for the more verbose generative AI responses.
“Dynamic Interaction is capable of creating incredible user experiences, and allowing consumers to interact with computers in the most natural, intuitive way,” SoundHound founder Keyvan Mohajer said. “By introducing Generative AI, we’re providing a gateway to a new world of ideas, information, and capabilities – from anywhere, completely hands-free.”
SoundHound is also upgrading the assistant by combining Dynamic Interaction with earlier features for the first time. That includes the domain-specific databases called Collective AI and its Deep Meaning Understanding technology to ensure the generative AI can still provide real-time information and produce the custom responses and actions programmed by the brand running the voice AI. SoundHound noted in an email Voicebot that the video shows the first demo of Dynamic Interaction for cars, following its initial setup for restaurants. The company didn’t disclose the details of the source of the generative AI except to say that they “can come from a mix of SoundHound and non-SoundHound models depending on the topic” and that “they are similar to the responses that ChatGPT would provide. “The company plans to share more about the generative AI element in the near future.
SoundHound’s rollout of new features and partnerships comes despite major shakeups in its operation. The company just raised $25 million in equity financing from an unknown set of investors, but only after reducing staff from around 450 to 200 people in two rounds of layoffs. With a goal of reaching profitability this year, SoundHound likely needs both the spending cuts and potential revenue increase represented by the hot field of generative AI. That said, SoundHound’s chief revenue officer Zubin Irani has left the company this month after just over a year. He’s back as CEO at consulting firm Cprime, a role he first left seven years ago. SoundHound’s senior vice president of people and culture, Angeline Tucker, also left in recent weeks for a similar role at mobile growth platform Digital Turbine after nearly two years at SoundHound.
Still, SoundHound does have a lot of new and expanding projects to highlight. Last quarter, SoundHound revealed new features for on-device and cloud voice AI and introduced a real-time transcription and annotation service. The company has also begun integrating its voice AI into the Harman Ignite Store for cars and released its first Mandarin language outlet as a feature for Dongfeng Peugeot Citroën Automobiles (DPCA) vehicles in China, contracted with Stellantis in Europe.