NLU component for Voice Assistant

Incorporating voice assistants into services and devices significantly improves the customer experience by implementing innovative functions and interactions. Today’s most innovative companies not only use algorithms that understand natural language, but have seen them become an indispensable ingredient of the voice assistants they offer as a supplement to their products and services.

deepsense.ai was tasked with developing a natural language understanding-based (NLU) solution for a voice assistant used by a global manufacturer of consumer electronics and smartphones.

deepsense.ai developed and now maintains an NLU module and various text processing routines. The state-of-the-art language understanding algorithms utilising neural networks, including LSTMs, Transformers, and BERT, are refined and trained for various tasks such as intent recognition, slot filling and named entity recognition (NER). All this heavy, behind-the-scenes processing paves the way for customers to effortlessly make calls, check the weather, find their way around, schedule meetings and dozens of other marvels they accomplish with their fingertips or, increasingly, their voices - and take entirely for granted.

We provide comprehensive support encompassing data quality assessment, model development and validation, and deployment, enabling our client to quickly and confidently release new functionalities and their customers to enjoy regular improvements and extensions of the voice assistant.