New Apple Machine Learning Issue: Finding Local Destinations with Siri’s Regionally Specific Language Models for Speech Recognition
In April Apple's Machine Learning Journal published Issue 9 titled "Personalized Hey Siri." Apple's Siri Speech Recognition Team have now updated their journal with Vol. 1, Issue 10, August 2018 titled "Finding Local Destinations with Siri's Regionally Specific Language Models for Speech Recognition."
The Problem: The accuracy of automatic speech recognition (ASR) systems has improved phenomenally over recent years, due to the widespread adoption of deep learning techniques. Performance improvements have, however, mainly been made in the recognition of general speech; whereas accurately recognizing named entities, like small local businesses, has remained a performance bottleneck.
The Solution: Apple describes how they met that challenge by improving Siri's ability to recognize names of local POIs by incorporating knowledge of the user's location into their speech recognition system. Customized language models that take the user's location into account are known as geolocation-based language models (Geo-LMs). These models enable Siri to better estimate the user's intended sequence of words by using not only the information provided by the acoustic model and a general LM (like in standard ASR) but also information about the POIs in the user's surroundings.
The issue is broken down into the following segments: Introduction; How Does Siri use Geo-LM; Geo Regions; Algorithm; Geo-LM Construction; Experiments and Results, Data; Results and Conclusion.
Read the full August 2018 issue here.
Figure 3. A simplified example of WFST-based class LM
About Making Comments on our Site: Patently Apple reserves the right to post, dismiss or edit any comments. Those using abusive language or negative behavior will result in being blacklisted on Disqus.