Network training for continuous speech recognition

The standard training approach for a hidden Markov model (HMM) based speech recognition system uses an expectation maximization (EM) based supervised training framework to estimate parameters. EM-based parameter estimation for speech recognition is performed using several complicated stages of iterative reestimation. These stages are heuristic in nature and prone to human error. This paper describes a new training recipe that reduces the complexity of the training process, while retaining the robustness of the EM-based supervised training framework. This paper show that the network training recipe can achieve comparable recognition performance to a traditional trainer while alleviating the need for complicated systems and training recipes for spoken language processing systems.