Faculty of Art

Drawing and Painting

Fiona Enright

Fragments of Speech Recognition Algorithms Research

"Speech recognition begins with digitizing the speech sounds and converting them into a standard, compact representation. The analysis can be based on matching the input sounds to one of about 200 “spectral equivalence classes” from which the representation can be created. Alternatively, algorithms can use data based on modeling how the human vocal tract produces speech sounds, and extract key features that then become the speech representation. Statistical approaches (such as Hidden Markov Models) are popular, and neural networks using various learning and recognition approaches can be “trained” to recognize speech features." Henderson, H. (2017). speech recognition and synthesis. In H. Henderson, Facts on File science library: Encyclopedia of computer science and technology (3rd ed.). Facts On File. Credo Reference: http://ocadu.idm.oclc.org/login?url=https://search.credoreference.com/content/entry/fofcomputer/speech_recognition_and_synthesis/0?institutionId=4079