Conversational & Generative AI
Conversational & Generative AI
Generative AI and large language models are making AI more visible and accessible than ever before, driving innovation and bringing AI-powered user experiences to the forefront. Cerence AI is leading the way in bringing AI to the car, with solutions that are transforming the way drivers and riders interact with their vehicles.
For our customers, this means flexible, fully branded, and customizable solutions that provide a unique balance of high-quality user experience, cost, and performance. For drivers, Cerence solutions enable intelligent, context-aware assistance that can handle a wide range of interactions, making the in-car experience more personalized, intuitive, engaging, fun, and productive.
Gen AI & CaLLM
Generative AI and large language models are transforming user experiences across industries. At Cerence AI, we are deeply focused on leveraging the latest in AI innovation to advance the future of the in-car experience. CaLLM™ (Cerence Automotive Large Language Model) and CaLLM™ Edge, our proprietary automotive large and small language models, leverage our extensive, fine-tuned automotive dataset to serve as the foundation for these next-generation user experiences.
CaLLM, powered by NVIDIA technology, has a unique level of automotive-specific intelligence, leveraging our extensive automotive expertise and dataset to deliver integrated in-car user experiences beyond those of general knowledge LLMs. CaLLM is distinctly able to support automotive functions, features, and requirements and is deeply customizable for automakers through training, fine-tuning, and bespoke applications. Its features include user personalization as well as built-in information retrieval for our generative AI-powered applications like Cerence Car Knowledge.
CaLLM Edge fits comfortably when embedded in the automotive headunit and is fine-tuned on Microsoft’s Phi-3 family of small language models, using our dataset to deliver highly specialized AI that can handle a variety of automotive use cases. Its core capabilities include implicit and explicit car control commands (for example, temperature, windows and doors, seat position) and point-of-interest search and navigation, as well as conversational interaction like, “What's the most popular movie ever shot in Hollywood?” followed by “Can you tell me more about the plot?”
Cerence Assistant
Cerence Assistant is our core offering, a turnkey in-car assistant platform that empowers automakers with an intuitive, AI-powered companion that can support drivers like never before. Cerence Assistant runs directly on a vehicle’s headunit, with advanced AI deeply integrated with car sensors and data to understand complex situations both inside the vehicle and around it. Serving as the central brain of the car, it flexibly and securely integrates edge technologies with cloud services to make driving more intuitive, connected, and enjoyable.
Cerence Assistant leverages advanced natural language understanding to understand individual preferences, personalize entertainment, and execute complex, contextual commands. It can also find preferred parking, play users’ favorite songs, direct them to the gas stations with the best prices, and prioritize the voices it needs to hear while muting the ones it doesn’t, maximizing the safety, productivity, and enjoyment of every customer journey.
Cerence Chat Pro
Consumers have become accustomed to asking their devices questions to get answers to a variety of topics – from weather, stocks and news to sports, entertainment and facts on just about any subject – whether at home or on-the-go. ChatGPT has only furthered this trend, bringing AI to the forefront of how consumers seek information. We’re bringing this same level of intelligence to in-car assistants with Cerence Chat Pro, the fastest and simplest way for OEMs to integrate LLMs, including ChatGPT, into their in-car assistants with full control over the user experience. OEMs can define their own persona, maintaining their brand identity throughout the experience, and change answers at any time via a self-serve web-portal, ensuring accuracy and safety.
As cars have become more complex, navigating the many features, notifications, and warning lights in vehicles has become increasingly difficult for drivers. Cerence Car Knowledge provides drivers with the best answers and information—all of which are tailored to each individual car, its configuration, and the continuously learned preferences of the driver—via a companion application, voice output from the automotive assistant, and imagery displayed on the car’s infotainment system. Now leveraging generative AI, Cerence Car Knowledge is more intelligent than ever.
Cerence Car Knowledge
Cerence Just Talk
Push-to-talk buttons have long been the standard ways that users begin their conversations with their assistants. As technology advances and voice assistants become more prevalent both inside and outside the car, we’re working with leading automakers and mobility OEMs to bring wake-up words to the car – think “Hey Mercedes.”
But what if your assistant could go a step further and simply knew you were talking to it – no button or “hey” required?
Enter Cerence Just Talk, AI-powered technology from Cerence AI that has the smarts to detect when drivers are talking to the assistant, and to keep quiet when they’re not. With Just Talk, drivers can simply begin speaking when they have a question or command for the assistant.
Human-Like Voice Output
As the use of in-car voice assistants explodes, they have become an innovative point of differentiation. Mobility OEMs are looking for solutions that can deliver the most natural and human-like voice output quality, unlocking new use cases for their users.
With our advanced, neural net-based text-to-speech (TTS) technology, we’re bringing a stunningly human-like virtual personality into the car, delivering the voice output quality that drivers expect in the expanding world of virtual assistants. Our most advanced TTS engine is nearly indistinguishable from a human voice, including natural pausing and breathing and varying tonalities based on content.