CODEWITHSUNDEEP
speech-recognitionhidden-markov-modelsstates
Rabia Saleem
Rabia Saleem

Reputation: 1

No. of states for HMM acoustic models in Speech Recognition

Is it necessary that hmm model for each word has same no. of states in speech recognition?

Upvotes: 0

Views: 178

Answers (1)

Nikolay Shmyrev
Nikolay Shmyrev

Reputation: 25220

No, in HTK you can choose arbitrary number of states for every model. In Kaldi you usually have 3 states in non-silence phones, 5 states in silence phones. In CMUSphinx each phone has the same number of states for simplicity.

Upvotes: 1

Related Questions