Re: Redefining [a] and [æ]? or using a better vowel quadrilateral?

On Aug 10, 10:55 am, Nathan Sanders <nsand...@xxxxxxxxxxxx> wrote:
In article
<5bb4d815-7eb6-4296-951d-590881e30...@xxxxxxxxxxxxxxxxxxxxxxxxxxx>,

"ranjit_math...@xxxxxxxxx" <ranjit_math...@xxxxxxxxx> wrote:
On Jul 29, 2:22 pm, Sonja Elen Kisa <sonj...@xxxxxxxxx> wrote:
Is Praat able to calculate F2-prime, or is the formula posted
anywhere? Is F2-prime debatable, or is it generally agreed upon as
more accurate than F2 to map a vowel?

Is there a way to define an F1' and F2' such that while differences in
F0 (pitch) between the following two lines in a hymn might have the
side effect of changing F1 and F2 too, it wouldn't change F1 prime and
F2 prime? (i.e., the "ee" in the two instances of "thee" would have
the same F1' and F2').

Just a closer walk with thee
Just a closer walk with thee

F1 and F2 are resonance frequencies, based on the shape of the mouth
(above the glottis). Assuming you hold your tongue, lips, etc. in the
same position for each instance of "thee", (the center frequencies of)
F1 and F2 won't change, even if the pitch changes.[1]

Of course, there will be minor variations in the formants, because
humans don't have infinitely precise control over their muscles. If
you want a single measure of F1 and F2 to represent this particular
speaker's pronunciation of [i], you just take the average.

And if you want a single measure for each formant *across* speakers,
you have to do some sort of normalization (e.g., multiplying a
speaker's formants by a scaling factor, like the ratio of the sum of
all of his formants to the sum of all formants from all speakers).

[1] This is not quite true. If the pitch is high enough, it could
exceed F1, in which case, there would be no F1. This is a known
problem in opera, and composers know to avoid putting high vowels
(vowels like [i], [y], [u] with the lowest F1, sometimes below 300 Hz)
on high notes, to avoid formant loss.

Hey ,how did you get access to my Acoustic Phonetics term paper from
1973?

(I even have a tape of a singer singing the five vowels at frequencies
100 Hz apart from however low she could go, and her little daughter
added some that were higher than she could go. Unfortunately we never
had the opportunity to randomize the noises and test them on subjects.
It's probably a lot easier now than it would have been with a razor
blade and splicing tape.)
.

Relevant Pages

  • Re: why cant the BBC
    ... Moreover, the Hz's of formants are the pitch of any actual tone, but ... at exactly the same pitch and the listener can still reliably tell ... speaker to make contrastive distinction between vowels. ...
    (sci.lang)
  • Re: Prosody-only utterance
    ... and the vowels are a neutral sound ... It amounts to with pitch contour roughly. ... > pair with a reversal of the order of the syllables, ...
    (sci.lang)
  • Re: OED pronunciation transcriptions
    ... Note that MWCD11 gives "duck" as having a schwa. ... British English makes the distinction. ... Ladefoged vowels" that calls into question the validity of his oral ... differences in the frequencies of the first and second formants. ...
    (alt.usage.english)
  • Re: How formants structures affect autocorrelation peaks detection?
    ... Did you see the pitch as regular repeats of the waveform in the ... The major Peak in the autocorrelation function corresponds to pitch ... harmonics in the power spectrum, ... formants are not regularly placed in the spectrum. ...
    (comp.speech.research)
  • Re: vocal pitch correction toolkits
    ... I've been looking for a free vocal pitch correction software or toolkit ... The results were promising: formants were ... My conclusion was that the final step should always be an LPC ... Rubberband does a fine job on pitch shifting, ...
    (rec.audio.tech)
Quantcast