Location-Dependent Vocabularies and Speaker Style Personalization for Accurate Mobile Diet Recognition

Obesity increases the risk of many health conditions and obesity rates have more than doubled since the 1970s [1]. Diet tracking is an important behavior change strategy for controlling obesity. However, manually recording foods eaten is tedious. Speech is one of the most natural methods of interaction. However, in practice, factors such as environmental noise and individual pronounciation styles result in low speech recognition accuracy. Our work focuses on accurately recognizing food orders as users order food at restaurants. We improve recognition accuracy using location-dependent speech recognizer vocabularies and speaker personalization.