This plugin implements De Looze and Hirst’s two-pass utterance-specific pitch threshold estimation algorithm. From the researcher’s website:
In order to avoid possible pitch tracking errors, pitch floor and pitch ceiling are set [to
q25 * 0.75
andq75 * 1.5
(whereq
stands for “percentile”), which] have been shown123 to give a better estimation of pitch extrema, i.e. to exclude more octave errors or microprosodic effects at the extreme of the f0 distribution, than setting pitch floor and ceiling parameters to the default values [in Praat] or to default values depending on the gender of the speaker.
The scripts and procedures provided by this plugin are written so that their main intended objective is the creation of Pitch objects with the estimated pitch thresholds. In writing them, scripts provided by the researchers in their website and in the Praat user’s list were taken into consideration.
This script takes a selection of Sound objects, and applies the algorithm described above to each in turn to generate a corresponding Pitch object. The resulting objects are selected at the end.
The parameters passed to the script are the floor
and ceiling
factors that
will be multiplied with the first and third quartiles respectively to estimate
the pitch thresholds. A floor factor of 0.75
and a ceiling factor of 1.5
are
recommended by the researchers in most cases, but they warn that for “expressive
speech”, a ceiling factor of up to 2.5
might give better results.
Passing zeroes instead of these values uses the recommended default values instead.
Similar to To Pitch (two-pass)…, this script takes floor and
ceiling factors, and an input and an output path. Designed for batch processing,
this script processes all the sound files in the input$
directory and saves
the resulting Pitch objects in the output$
directory.
Passing zeroes as the factors uses the recommended default values.
sound = selected("Sound")
@pitchTwoPass: 0, 0
assert numberOfSelected("Pitch") == 1
This procedure has the same effect as the To Pitch (two-pass)… script described above, and works in the same way. In fact, it is used internally by all other scripts.
It takes a selected Sound object and generates a Pitch object, which will be
selected after completion. For ease, the ID of that object is also stored in the
.id
variable.
De Looze, C. and Hirst, D. (2008) Detecting changes in key and range for the automatic modelling and coding of intonation, Speech Prosody. ↩
De Looze, C. (2010) Analyse et interprétation de l’empan temporel des variations prosodiques en français et en anglais contemporain, Doctoral thesis, Université de Provence. ↩
Hirst, D. (2011) The analysis by synthesis of speech melody: from data to models, Journal of Speech Sciences 1(1): 55-83. ↩