At the beginning of March 2024, DS Automobiles became the first automaker in the world to go into full production with SoundHound Chat AI, with an initial rollout in 13 languages across 18 countries. The most sophisticated voice AI assistant available today, it can respond to a vast range of questions from hundreds of real-time domains, as well as large language models, like ChatGPT.
This assistant – which DS Automobiles has named Iris – will allow drivers and passengers to use hands-free voice control to unlock a vast range of information and updates. Whether that’s for planning a trip, opening the sunroof, asking for the sports scores, making a call, or accessing parking information from their destination.
Alongside familiar requests, like navigation, texting, and weather updates, voice assistants with integrated generative AI help drivers by answering questions like:
- What’s the typical weather in Tokyo in March?
- Tell me about the best museums there?
- Navigate me to the museum you just recommended
- What foods are seasonal in Japan during the springtime?
- Where can I eat those dishes?
The SoundHound system also uses a proprietary approach to massively reduce the risk of “AI hallucinations” – misleading and unpredictable responses often associated with some large language models (LLMs).
“We are excited to see our advanced voice AI enter the Japanese market,” said Mike Zagorsek, COO of SoundHound AI. “This technology is the first of its kind to go into production. That’s because innovative automotive brands like DS Automobiles continue to demonstrate a commitment to building the vehicle experience of the future.”
First data shows huge increase in usage
SoundHound AI has now piloted this technology across a range of high profile OEM brands, and has found that adding generative AI capabilities to in-vehicle voice assistants hugely increases the frequency with which drivers and passengers use them – with DS Automobiles reporting an increase of over 50%.
This enthusiasm for ChatGPT-like functionality bears out the findings of SoundHound AI’s 2023 driver survey, where 50% of all regular drivers responded to say they were likely to use it.
How SoundHound Chat AI works
SoundHound Chat AI platform can integrate with any third party generative AI model to create a seamless conversational experience. It keeps the conversation going, providing fast, accurate responses to queries without frustrating results (no more “I’m 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.
With the launch of SoundHound Chat AI, SoundHound unveiled its proprietary approach, which uses software engineering technology named CaiLAN (Conversational AI Language) and machine learning technology named CaiNet (Conversational AI Network) to ensure fast, accurate, and appropriate responses. The former, deploys software engineering to efficiently build knowledge domains (like weather, restaurants, traffic, and local search), while the latter uses machine learning to better understand queries and provide the right responses. CaiNET can connect to SoundHound’s internal models as well as external models such as OpenAI’s GPT, among others. CaiLAN controls and arbitrates the results to provide the best response to the user.