I. Getting Started on XVocal
In order to run the program, you have to set the path in your .cshrc file
setenv LD_LIBRARY_PATH /usr/local/X11R5/motif/lib:$LD_LIBRARY_PATH
II. Run the program
Run the executable file
(For better color, you may try
xvocal -background white)
III. Main Menu
There are six items in the main menu:
- VT Calculation
Provides a choice of five different vocal tract models for calculation.
Change the physical configuration of the vocal tract.
- Physical Constants
Modify the physical constants in simulation
- Save Option
Modify the save options
Quit the program
IV. VT Calculation Menu
There are five models of vocal tract available for simulation:
After choosing a model, a menu will be poped up on the screen. There are three
common items in the menus:
- Uniform tube model
Area function is specified by the uniform cross-sectional area, the tube length
and the number of sections.
- 2 Tube model
Area function is specified by
1. Cross-sectional area of the 1st section
2. Length of the 1st section
3. Area of the 2nd section
4. Length of the 2nd section
- 3 Parameter model
Area function is specified by Fant's 4 tube model with 3 parameters:
1. Location of the tongue-body center
2. Cross-sectional area of the tongue-body section
3. Mouth-opening area
(Note in the original model that the 3rd parameter is the length over area
ratio. But we assume, here, the lip tube length is fixed to 1cm. Ref:
Cunnar Fant (1960). Acoustic Theory of Speech Production, Mouton, The Hague,
- Linear Articulatory model
Area function is specified by parameters
- From file
Area function is read from file in specific format.
The data files, with extension .are, are in ASCII in the following format
no. of sections
length of a section
cross-sectional areas of each section (from glottis to mouth) separated by
Sample files can be found in AREA directory.
Other items in the menus are just parameters for modification. Simply click on
the item and a prompt dialog will be poped up for you to enter the new value.
Compute the frequency spectrum and the formant frequencies.
Synthesize a stationary vowel with the current area function. The duration is
about 300ms, and a predefined rise-fall F0 pattern is used. The vowel signal
and glottal signal is stored in temporary files. To transfer onto a permanent
file, select menu "save".
- Keep (Save)
Save the current tract calculation conditions in a text file. The file name
consists of the save_file_code followed by the current value of the file_count.
The catalogue code is .DOC for area function and .SIG for synthesized signal.
V. Configuration Menu
Four parameters of the vocal tract can be modified here:
The configuration can be changed simply by clicking the item to toggle between
on and off.
- Radiation boundary: RL-circuit of short circuit, at the lip or nostril
When toggle button is on, there is radiation and set to RL-circuit.
- Wall characteristics: Yielding or rigid
When toggle button is on, there is wall yielding.
- Nasal tract: On or Off
When toggle button is on, nasal tract is on. Nasal coupling is further
specified when you choose a model from VT Calculation menu again.
- Glottis: Open or Close
When toggle button is on, glottis is open.
(At this moment, spectrum is assumed to be empty when the glottis is open
VI. Physical Constant Menu
The value of the following physical constants can be changed by clicking on the
item and entering new value in the prompt dialog.
|Physical constants||Default Values|
|Air Density||0.00114 gm/cm3|
|Sound Velocity||35000 cm/s|
|Wall Resistance||1600 gm/s/cm2|
|Wall Mass||1.5 gm/cm2|
|Wall Compliance||300000 gm/cm3|
VII. Save Option Menu
Filename consists of a fixed code followed by an integer, which is
incremented every time a save file is created. The fixed code can be changed by
Save file code item. With the new code, the counter
To modify any item, just click on it and enter the new path or file code in the
Save directory option changes the directory of the
synthesized file to be saved. Default is /SIG/.
temp_glt_path are the temporary
file names for synthesis of stationary vowels.
Area directory option changes the directory of the area
function files. Default is /AREA/.
VIII. Vowel Playback
To playback the synthesized vowel(see IV. VT Calculation Menu 2. Synthesis),
you have to exit XVocal. Then change to
SIG directory. The
files are the binary files of the vowel. You may convert it to
any format for any audio players.
Some Matlab programs were developed by Edmond
Chui to play it back in Matlab and convert the binary file to
format. You may try one of the following:
You may then exit matlab and play the .au file using audiotools.
- To play the signal in matlab, first use the function convert to downsample
the 10 kHz signal from XVocal to 8 kHz. Then, use the sound function to play it
back. The syntax is as follows:
>> y = convert(filename); % filename must be in single quotes
% and with extension .SIG
>> sound(y, 8000); % must be 8000 for the sampling rate
- To convert the signal to .au format and play it outside matlab, do the
>> play(in_file, out_file); % in_file must be .SIG and
% out_file must be .au
Back to XVocal home page