AI in the Era of Voice Interfaces and LLM: From Embedded to Cloud

Large language models like ChatGPT may have been dominating your professional social media feed since the end of 2022. With their large footprint, they are an example of cloud-deployed AI that may be a part of a larger system. At the other end of the spectrum, AI models are increasingly entering devices around us through their small footprint embedded implementations. Speech interfaces in voice-enabled devices are an example of systems spanning the whole embedded-to-cloud implementation spectrum, from audio preprocessing and wake-up phrase detection to spoken language understanding. When it comes to AI-powered engineered systems, you can use MATLAB^® from creating the training data to the final deployment. You can also adopt it for building early prototypes that may include external hardware, embedded code, other programming languages like Python^®, cloud services, and custom app interfaces. In this session, you’ll learn how to create a real-time prototype of a voice interface complete with embeddable AI models built in MATLAB, deep networks developed with PyTorch^®, and the use of ChatGPT for language understanding. You will also explore other engineering workflows involving the use of LLMs with MATLAB.