Auditory analysis of speech sounds: theory and practical training (2008)

 

The class is organized by

Date and place

 

Prerequisites

The course is aimed at doctoral students and master students with a basic knowledge of speech technology (the equivalent of a GSLT level 1 course in Speech technology).

Overview

The course is obligatory for all our post-graduate students on Phonetics and Speech technology. The course will give 4 ECTS credits

The aim of the course is to provide the basics of auditory analysis of speech and to introduce the participants to various aspects of analysis of sounds of different languages.

The course is intended for those who are interested in auditory analysis of the sound systems and who would like to get a practical training in the application of the IPA to phonetic analysis.

Lectures on the introduction to phonetic description of speech sounds will be followed by a tutorial on auditory analysis, phonetic description and transcription of speech sounds with the use of the IPA notation system. In order to present the realization of the speech sound in its relation to its perception and notation an acoustic display of the speech signal will be used.

The application of the IPA will be demonstrated by the sound systems of different languages: English, Russian which have been studied and analyzed extensively. The analysis will also include the Avar and Karata languages (the Caucasian languages of Dagestan). The sound system of Avar has been studied less extensively and the phonological system of the Karata language has not been described at all. Native speakers of Russian, Avar and Karata languages will participate in the seminar.

Problems arising in the perception and interpretation of sounds as "similar" or "different" by speakers of different languages will be discussed.

Where to use the information and skills:

-- in collecting data for a speech database to be used in concatenative (or other) speech synthesis, since in preparation of the segment inventory phonetic speech material is to be phonetically labeled, and a transcription (broad or narrow depending on the requirements and the qualification of the labeler) is done by EAR. Which means that
transcribers should be well trained and experienced to be able to analyze (from the point of view of articulation and acoustic information), describe and properly annotate (using IPA or other notation system) real speech events, like contextual variants of phonemes, inserted and elided sounds, vowel and consonant positional length, voicing and devoicing processes, assimilation of speech sounds in real speech, which may not be described by general allophone rules, etc.
Labeling is a laborious and time-consuming task, since it is done manually and controlled by ear. Its quality and progress depends on the experience of the labelers.

-- in the system evaluation: perceptually significant differences between realizations of speech stimuli are determined by listening tests.

-- in collecting data for a speech database to be used in ASR systems.

-- in experiments with physical modification and generation of sounds.

 

Format: Lectures and exercises

Time slots for lectures:

Monday October 13th: 11:00-12:30, 13:00-14:30

Tuesday October 14th: 11:00-12:30, 13:00-14:30

Wednesday October 15th: 11:00-12:30, 13:00-14:30

Thursday October 16th: 11:00-12:30, 13:00-14:30

Friday October 17th: 11:00-12:30, 13:00-14:30

ECTS credits: 4

Assessment: pass/fail grade



Avar-plosives