I'm currently labelling some singing data, which contains very slow transitions between vowels. The data often have prenuclear glides and diphthongs together. The picture below is an instance (the correct IPA should be [ti̯ɑʊ̯]):


If I understand correctly, from the /i/ sound to /a/ has a transition with the shape of a ">", since /i/ has lower F1 and higher F2 than /a/; from /a/ to /u/, the shape is the other way round.

However, the descending or ascending slopes are just the transition periods. How do you decide where to put the boundaries? Are there textbooks or articles discussing such issues? Should I put the boundaries in the middle of the transitions?

I am aware of a similar question here: Identifying phonemic boundaries in Praat

However, the answers in that question did not answer my questions; and I'm adding a practical example to see what other linguists would do under similar circumstances.

  • 2
    Possible duplicate of Identifying phonemic boundaries in Praat
    – Nardog
    Jun 5, 2019 at 6:50
  • It's not useful to denote a precise start and end point, is it? You rather denote articulatory features in the signal, so you'd better ask a trained singer. Also, middle is not a precise term, because different measures exist (arithmetic mean over time, power, frequency; median; I don't know, I'd wonder what signalprocessing.se would say), which implies that it doesn't make a difference as long as it is consistent.
    – vectory
    Jun 5, 2019 at 17:41
  • @vectory why did you say it's not useful to denote the precise start and end points? I was trying to figure out some durational patterns of the phonemes, so I need to make some decisions. I understand that it's impossible to find the precise points but i was wondering whether there was any conventional ways of dealing with such cases. Like how did people who study durations in connected speech decide where to put their boundaries? I haven't really found any paper that has such details. Jun 6, 2019 at 6:34
  • It depends a lot on the purpose of the segmentation. Obviously, singing is not the same as speech...
    – user23078
    Jun 7, 2019 at 12:11

1 Answer 1


Only a proposal:

The problem seems to be: having a clear definite point to determine whether you are in one vocal or another for example.

As you said you could define transition zones. And then take a temporal "barycenter" of those zones as the discrete transition points so to base your calculation of durations.

My point is: You can get some inspiration from Fuzzy Logic, which helps a lot with the need to describe a quality / a feature that is not a discrete value. I think it is exactly the case for phonems.

As you highlighted with your varying frequency curves, and when looking at the IPA diagram, it is clear that the space between vocals is totally continuous, and not discrete:

enter image description here

Same for consonants, I don't have a diagram at hand, but when playing didjeridoo/yidaki, I spend my time exploring the nuances between all sorts of d from the back of the palate, to the d on the front of the palate, then approaching the different t and d we can do on the back of teeth, either at the base of the tooth/gengiva or at the top of the teeth.

Same for k and g. You have all the palette of tongue position, and strenght in the breath, that makes them sound slightly different.

And for ch, f, v, and h, the palette is continous also.

Like you have asked, the difficulty is determining the spectral/frequency features to define one phonem. You are asking for reference about that, I don't have any. Neither have I used Praat.

But you can start arbitrarily fixing them. And decide yourself: when this or that parameter has gone to far away from this value, we are no longer in this phoneme. When this parameter and this parameter is there: we are totally in this phonem.

That's the core of Fuzzy logic: only defining the points "We are full in it", "We are no longer in it".

Taking an ultra simplisitc example from the desciption of Fuzzy Logic with only one dimension / degree of liberty, (here it's temperature) you happen to have something like this:

  • 0 : Quality is false
  • 1 : Quality is true, fully present

enter image description here

It works really well for multi-dimension continous space: that's namely the strength of this technique indeed. And I tend to think you could apply that to several parameters for defining phonems.

As said, it's only a proposal. I had worked on a simplistic system which was determining at which moment the sound was a vocal or a consonant. And it was mostly based only on these two qualities:

  • noisy
  • pseudo-periodic

I was not using formant frequencies like you seem to do (that's what I interpreted from the graph).

Ultimately, if the description of each phonem you are interested in is too tedious, you could think about using Machine Learning. You would have to provide examples of frequency analysis (not the sheer impulse amplitude data which is too precise, detailing too much, too much local) and the corresponding phonem. That could be a good candidate for what is called "Classification" in machine learning. But you would have to invest in learning some ML classification techniques.

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.