Workshop Analysis and Synthesis of Speech

Week scheme

1. Making and copying an analogue sound recording.

The principles of a microphone, headphone and a loudspeaker. Theory and practice of analogue sound recordings. The usage of a function generator. Wiring up equipment.

2. The computer as a signal processor.

Speech and acoustics. The basics of digitizing signals. How to interpret an oscillogram of a speech signal. How to segment and label speech. Fundamental speech manipulation in the time domain using the Praat software.

3. How to operate a PC with MicroSoft Windows.

Simple command line commands compared to the usage of the Graphical User Interface. Smoothly working with files and directories, creation of lists with file names. Organizing and automating various procedures. How to find more information. Basic rules in Information Management.

4. Making and processing digital sound recordings.

How to make a good quality sound recording on a digital medium. Working with subjects. Some standards for digital sound recordings. Conversion of data between different media types (analogue tape, solid state recorder, CD, hard disk) and different formats (sample frequency, number of channels, compression).

5. Making and interpreting spectra.

A noise generator, an active filter and the HP 3582A Spectrum Analyzer compared to their equivalents in software. Understanding (inverse) Fourier Transformations. How to make a vowel diagram.

6. Making and interpreting spectrograms.

Making and understanding a spectrogram, using either the Sonagraph or Praat. How to prepare pictures and graphs to be printed on paper or for use on a computer screen.

7. LPC- en PSOLA analysis and resynthesis. Duration manipulation using Praat.

The speech synthesis model of Fant. LPC analysis en LPC (re)synthesis of speech. How to manipulate the temporal organization of synthesized speech.

8. Pitch manipulation using a PitchTier and a Praat script.

Analyzing, interpreting and modifying the pitch parameter in speech with the PSOLA technique.

8a. LPC Formant(re)synthesis using Praat-scripts.

How to design and implement a number of scripts to increase the possibilities of Praat.

9. Advanced usage of MicroSoft Windows and its applications.

Adjusting some of the many settings in MS Windows to the needs of a phonetician. How to install Praat scripts to increase efficiency when working with large amounts of data. Writing and executing (interactive) batch procedures and Praat scripts. Setting up a speech database.

10. Selecting and combining data in different TextTiers.

How to collect, select and organize results of measurements using a simple scripting language like AWK. Exporting/importing data between Praat and other software like SPSS. How to use and modify published AWK snippets.

11. Plotting unstructured data using Praat.

Displaying phonetic data in 'non-standard graphs'. In other words: How to use Praat to draw arbitrary data.

12-i. Other signal processing software.

Using the speech analysis and - manipulation program Gipos, to visualize the relations between an oscillogram, a spectrum and a spectrogram. How to use Adobe Audition to process multi-channel recordings.

12-ii. Using E-Prime.

The design and implementation of a Phonetic Reaction Time experiment using the visual programming environment E-studio. Processing collected data with E-DataAid.

12-iii. Speech synthesis with Fluent Dutch Text-to-Speech software.

Creating artificial speech utterances, meeting strict conditions.

13. Collecting subject data using basic Internet techniques.

Using networked computer systems. How to design, implement and run a simple Phonetic Perception Experiment over the Internet, using public domain languages (HTML, PHP) and software (Web-browser, FileZilla, E-Mail and HTTP-server).