Firefly algorithm for log-likelihood optimization problem on speech recognition

Speech Recognition System is a system that listens human speech and compares the speech with words or phrases that have been prepared in advance in order to obtain the data. To optimize the model we employ fireflies to search λ = (A, B, π) that produce maximum log-likelihood. We evaluate the performance we compare the technique with PSO. Firefly algorithm outperforms PSO on constructing HMM for speech recognition. FA performance can be seen on the log-likelihood values which are larger than PSO. FA guesses more words correctly indicating higher accuracy as well.