Course details

Speech Signal Processing for Speaker Recognition

QRH Acad. year 2002/2003 Winter semester

Current academic year

Phonetic description of the Czech, English and German language, speech production and perception by humans, calculation of speech parameters, statistical methods of voice processing, definition of various distance measures, decision criteria, speaker verification (passwords), speaker identification (unknown phone calls), estimation of psychical state of the speaker, noisy speech enhancement, creation of special databases, available speaker verification systems, some typical applications of speaker recognition.

Guarantor

Sigmund Milan, prof. Ing., CSc. (UREL)

Language of instruction

Czech

Completion

Examination

Time span

  • 26 hrs lectures
  • 13 hrs pc labs

Department

Department of Radio Electronics (UREL)

Subject specific learning outcomes and competences

The students become familiar with the phonetic description of the relevant languages, speech signal features, speakers recognition approaches and some typical applications.

Learning objectives

The aim of the course is to make students familiar with the advanced methods for speaker recognition based on their voice analysis.

Prerequisite knowledge and skills

There are no prerequisites

Study literature

  • SIGMUND,M. Analýza řečových signálů. Skriptum FEI VUT, Brno 2000
  • PSUTKA,J. Komunikace s počítačem mluvenou řečí. Academia, Praha 1995
  • RABINER,R., JUANG,B.H. Fundamentals of Speech Recognition. Prentice Hall, Englewood Cliffs, N.J., 1993

Fundamental literature

  • PSUTKA,J. Komunikace s počítačem mluvenou řečí. Academia, Praha 1995
  • KLEVANS,R.L., RODMAN,R.D. Voice Recognition. Artech House, London 1997
  • BAKER,R.J., ORLIKOFF,R.F. Clinical Measurement of Speech and Voice. Singular Publishing Group, New York 2000

Syllabus of lectures

  • Introduction, theory of speech production.
  • Phonetic description of Czech, English and German language.
  • Speech features for speaker recognition.
  • Purport and estimation of fundamental speech frequency.
  • Automatic recognition of man/women/child voices.
  • Statistical methods of voice analysis.
  • Text-dependent voice recognition.
  • Text-independent voice recognition.
  • Speaker verification and identification.
  • Effects of emotional stress on voice.
  • Speech enhancement approaches.
  • Analysis of non-speech vocalisations.
  • Available speaker verification systems and software tools.

Syllabus of computer exercises

  • Speech visualisation methods, illustration of various voice types.
  • Calculation of selected speech parameters.
  • Measuring of fundamental frequency, voice jitter and shimmer.
  • Hidden Markov models based speaker verification.
  • Estimation of emotional stress using voice analysis.
  • Speech signal enhancement in noisy environment.
  • Analysis of an unknown voice (Test).

Progress assessment

Study evaluation is based on marks obtained for specified items. Minimimum number of marks to pass is 50.

Controlled instruction

There are no checked study.

Back to top