To get a picture of the ASR accuracies for individual phonemes for the two age groups, we trained monophone models, each phoneme modeled as a three-state HMM with 18 Gaussian components per state.
This is done by modelling each phoneme as a small HMM (Figure 2) and combining them into a larger HMM, such as the one shown in Figure 3, with a set of parallel chains such that each chain maps to one word; for example, given that we are in the state at some time, then we are also definitely (i.e., with probability 1) in Word A and Phoneme B at time.
Each expert is a classification model that employs one hidden Markov model for each phoneme.
The Markov-based approaches estimate a model for each phoneme and use n-gram models to compute(P(F)).
For phoneme-level training, the adaptation of each phoneme model was performed in two steps.
In [8],[9], the increased phonatory variability associated with dysarthric speech was addressed by a system enabling more suitable hidden Markov model (HMM) topologies for each phoneme in the speaker's repertoire.
A normal hidden Markov model is constructed in the way shown in Figure 5, with each phoneme being modelled as a mixture of expert models.
Statistical speech models and a probabilistic technique called Gaussian mixture modelling are then used to identify each phoneme, before reconstructing the original word.
If we have a sequence of phonemes, this is similar to training models, one for each phoneme-phoneme transition in the sequence, during synthesis (i.e., not as a preprocessing step).
The model defines a simple loop of phonemes, where each phoneme is a left-to-right three-state HMM.
