I am studying the acoustic correlates of breathy voice. Two of the measures I use are the difference between the amplitude of the first harmonic and the amplitude of a harmonic near the frequency of the first and third formant, respectively. While they make sense for stable monophthongs, they are less suitable for diphthongs and vowels near glides and liquids, where there is a long transition phase. What measures should I use in such cases? (just to be clear, I'm interested in the full segment, not just the stable edges)
It depends on a number of factors (incl. preceding/following segmental context), but as a rule of thumb, I'd suggest starting with two measurements: 1) the nucleus and 2) the offglide and compare the two using F1 and F2 vals. You can compare measurement 1) with the monophthongs as a first-pass check, assuming a comparative analysis is of interest.
Otherwise, bearing in mind the correlations between the tendency for breathy vowels to raise and back, it may be worth focusing on breathyness and onglides. One would expect breathy onglides to have lower formants (as a result of the lowered larynx and elongated vocal tract lowering F1 formants and resulting in higher vowels) than non-breathy or creaky onglides.
Hope this helps -
Does this one here help?
http://jslhr.pubs.asha.org/article.aspx?articleid=1780610
Acoustic Correlates of Breathy Vocal Quality
It is also on Researchgate. They mention signal periodicity as a main acoustic correlate. Could also be interesting to look at different acoustic correlates as they evolve in time as a time series, since different vowels have different properties and therefore everything may change during the transition.
Best Susanne
1. There is a follow-up to this paper that extends the methods to continuous speech rather than the sustained vowels that were used in the 1994 paper:
Hillenbrand, J.M., and Houde, R.A. (1996). “Acoustic characteristics of breathy vocal quality: Dysphonic voices and continuous speech,” Journal of Speech and Hearing Research, 39, 311-321.
The methods are described in somewhat more detail.
2. Software for the CPP-S measure that is described in this paper can be downloaded here:
http://homepages.wmich.edu/~hillenbr/
Look for these links:
cpps.exe (right click and choose "Save target as"); Readme file
The 'readme' file gives the basics for using the software.
http://homepages.wmich.edu/~hillenbr/
Thank you very much for all your answers and links!
I will try all the methods described in the 1996 paper. I already tried CPP and harmonics-to-noise ratio, but they don't seem to correlate well with the perceived breathiness of my data.
Try using VoiceSauce to extract the data you need : http://www.phonetics.ucla.edu/voicesauce/documentation/index.html
- Badges
- Science topic
- Similar topics
- Geoscience
- Cartography
- Mapping