Titlebook: Discriminative Learning for Speech Recognition; Theory and Practice Xiaodong He,Li Deng Book 2008 Springer Nature Switzerland AG 2008

廣口瓶

愛(ài)了嗎

Discriminative Learning: A Unified objective Function,HMMs). These are: maximum mutual information (MMI), minimum classification error (MCE), and minimum phone error/minimum word error (MPE/MWE). We also compare our unified form of these objective functions with another popular unified form in the literature.

meretricious

Discriminative Learning Algorithm for Exponential-Family Distributions,design where each class is characterized by an exponential-family distribution discussed in Chapter 1. The next chapter extends the results here into the more difficult but practically more useful case of hidden Markov models (HMMs).

Proclaim

FAR

Subdue

1932-121X ech recognition. The specific models treated in depth include the widely used exponential-family distributions and the hidden Markov model. A detailed study is presented on unifying the common objective functions for discriminative learning in speech recognition, namely maximum mutual information (M

星球的光亮度

CSR, Sustainability, Ethics & Governancey, real-world speech recognition tasks such as commercial telephony large-vocabulary ASR (LV-ASR) applications. We show that the GT-based discriminative training gives superior performance over the conventional maximum likelihood (ML)-based training method.

啤酒

NUL

Book 2008ition. The specific models treated in depth include the widely used exponential-family distributions and the hidden Markov model. A detailed study is presented on unifying the common objective functions for discriminative learning in speech recognition, namely maximum mutual information (MMI), minim

感染