HMM-based ASR

ASR based on comparing incoming speech with statistical models for speech units (acoustic models), usually in the context of a language model which deploys statistics about word sequences. 

• ·        Kai-Fu Lee is one of the best-known names in the development of HMM-based ASR. He recorded this lecture in 1993: 
•       There are two modules in Edinburgh's speech.zone: here and here
• ·        Hermann Ney: 2012 Review
• ·       Columbia Speech Recognition course Lectures 3-12
• ·       Andrew Maas: April 10 and 12 lectures