Large-vocabulary speaker-independent continuous speech recognition using HMM

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

Hacking smart machines with smarter ones: How to extract meaningful data from machine learning classifiers (2013, methodology): “The predominant technology used to perform this task is the HMM [37], many tools are nowadays available ([5, 41]).”
Subspace distribution clustering hidden Markov model (2001, background): “Phones (context-independent phones [10], generalized biphones/triphones [11]), states (tied-state HMM [12], [13]), observation distributions (tied-mixture/semicontinuous HMM [14]–[17]), and feature parameters [18] have all been tied.”
Gram-CTC: Automatic Unit Selection and Target Decomposition for Sequence Labelling (2017, methodology): “…word-pieces (Wu et al., 2016b; Collobert et al., 2016; Zweig et al., 2016a), words (Soltau et al., 2016; Sébastien et al., 2015), and phonemes (Lee and Hon, 1988; Sercu and Goel, 2016; Xiong et al., 2016b) have been widely used as basic units for text prediction, but all of them have…”
No NAT'd User Left Behind: Fingerprinting Users behind NAT from NetFlow Records Alone (2014, methodology): “Hidden Markov Models (HMMs) are widely used in sequences analysis since there exist efﬁcient algorithms to solve the three problems deﬁned above [9], [13], [18].”