The course teaches how the human voice is realised acoustically and how to analyse the acoustics digitally.  It also covers the evaluation of forensic evidence using forensic speaker identification as a model.

Topics:

What is a voice?

Phonetic Quality and Voice Quality

Tape recorders and recording informants

Basic Acoustic Concepts I):

Sound and its propagation

Dimensions and formula of sine wave

Basic Acoustic Concepts II): Types of speech wave.

Fourier analysis

Frequency- and time-domain representations. Filters.

Analog to Digital conversion (theory)

Analog to Digital conversion

Source-Filter Theory I)

Segmentation of the Speech wave I)

Digital signal processing

Spectrograms I)

types of Digital spectroGRAMS.

Making Digital Spectrograms

Measurement

Statistics  Distributions, Probability, T test. ANOVA, Bayes' Theorem

Voice onset time

Vowels I) Traditional articulatory theory; articulatory modeling;

acoustic plots and distinctive features; perception.

normalisation

Suprasegmentals:

Fundamental frequency (F0);

intrinsic F0

Digital F0 extraction 

tone;  tonogenesis.

Tone Perception

intonation;  multidimensional scaling analysis

F0 Normalisation.

(Forensic) Speaker Identification

Bayesian Inference and the Likelihood ratio

One written test (during one lecture)

One major research topic - each student conducts research into a suitable prearranged topic, decided either by the lecturer or the student.  Results of the research are reported orally and discussed by the group in the class presentation (one hour per student, use PowerPoint [if you can't I'll show you how]), and then written up. The research question can address either a Linguistic-Phonetic question (how Languages vary in their phonetics) or a Forensic-Phonetic question (how the voices of different individuals vary and the extent to which they can be discriminated.)

Weighting:

Research topic.........................   40%

Test....................................        10%

4 assignments ........................    50%