Signal and symbol phonetics (6 May 1999)

The following Web research tasks will provide you with basic information for the main specialisations which have already been discussed in class: child language acquisition, speech pathology, and automatic language recognition.

The task is to answer the following questions, searching for definitions and descriptions for the technical terms on the Web. You can start with the Reference Materials page, but also use search engines.

Try to structure the results of your research so that a multimedia Web page can be constructed using them:

1. Locate and print yourselves copies of the IPA.
2. Phonetic alphabets: Which phonetic alphabets can you find, and how do they differ? Examples: IPA, SAMPA, ...
3. ``The International Phonetic Alphabet is not just a convenient aid to writing down speech sounds -- it is actually a theory of speech segmentation and classification.'' Discuss this claim. If this claim is true, is the IPA the best possible theory of this domain? If not, why not?
4. Using the IPA, give transcriptions of the following sentence with several different German and English accents: Good morning! I think it's going to rain today.
5. How are the following defined/described? --
   • oscillogramme
   • spectrogramme
   • signal amplitude
   • signal energy
   • frequency
   • fundamental frequency
   • harmonic frequency
   • phase
   • wavelength
   • formant
   • formant diagramme (vowels)
   • burst
   • pitch perturbation (consonants)
   • intrinsic pitch (vowels)
   • microprosody
6. Using the CoolEdit signal editor, and speech files, make a small collection of different speech sounds (stops, fricatives, liquids, nasals, glides, vowels) and describe their acoustic structure.
7. Find definitions of `signal' and `symbol'. Why is acoustic phonetics `signal phonetics' and IPA transcription `symbol phonetics'?

Further hints: What are the following, and how do they work? --

1. Fourier series:
   • fourier series
   • fourier analysis
   • fourier transform
   • discrete fourier transform
   • fast fourier transform
2. Filters:
   • low pass filter
   • high pass filter
   • band pass filter
   • convolution (German: Faltung)
3. Digitising (for web search, try also the `ize' spelling):
   • discrete vs. continuous signal
   • discretising
   • sampling rate
   • sampling theorem (Nykvist theorem)
   • anti-aliasing filter
4. Pitch extraction:
   • zero crossing
   • peak picking
   • comb filter
   • cepstrum
   • autocorrelation
   • octave error