dear users,
I) I had downloaded the msys_mingw8.
+ source code of PRAAT "sources_5112".
II) I copied the sources_5112\makefiles\makefile.defs.mingw
to sources_5112\ ; and rename it to "makefile.defs".
III) Then I run "msys" and change the path to the "sources_5112"
then write "makefile".
--- the processing is done but with many errors.
--- no file get outputted...
where I am wrong in compiling the source code ??
Please help??
thanks
Amanpal Singh
At 06:32 -0800 19/11/09, shweta sinha wrote:
> I have calculated MFCC using pratt.How do I save the matrix
>obtained into a file so that can be distributed to other system.
You can save your MFCC object to a text file with "Write to text
file..." from the Write menu.
If the result is not in a format that you like, you could do "To
Matrix" first, and save the resulting Matrix with "Write to matrix
text file..." from the Write menu. If you want to interchange the
rows with the columns, do "Transpose" before saving the Matrix.
--
Paul Boersma
Phonetic Sciences, University of Amsterdam
Spuistraat 210, room 303
1012VT Amsterdam, The Netherlands
http://www.fon.hum.uva.nl/paul/
phone +31-20-5252385
Dear all,
as someone who regularly uses Praat under different platforms (Windows XP,
Ubuntu Linux, Mac OS X Leopard), I've become frustrated by one particular UI
issue under Mac.
That issue is click-through, or rather lack thereof. If I click into an
unfocused window, say to click a button, open a menu, make a selection,
whatever, and that actually happens *in addition to* giving the window focus,
that's called click-through.
In Praat on the Mac, however, I have to click *twice* to perform an action in an
unfocused window: once to give the window focus, then again for the action. And
since a typical Praat session involves numerous windows, and switching back and
forth between them, that makes for a lot of extra clicking.
On the other hand, this is a not an issue under Windows or Linux, where the
response to a click into a window never depends on whether or not that window
had focus. (In fact, non-Mac users will probably wonder what I'm talking about
in the first place.)
Under Mac OS X, click-through seems to be a feature of individual applications.
For example, XQuartz has a preference (in the Windows tab) to toggle this.
So my question is: has anyone else perceived this as an issue in Praat under Mac
OS X, and more to the point, is there a way to turn on click-through for Praat?
Best wishes,
-Ingmar Steiner
_______________
Phonetics/Phonology Section
Department of Computational Linguistics and Phonetics
Saarland University
Office: Building C7 2, Room 5.06
Email: steiner@...
Phone: +49-681-302-4696
http://www.coli.uni-saarland.de/~steiner/
Can I calculate both fundamental frequency (F0) and the pitch using praat? If
yes, how?
I understand that F0 corresponds to number of glottal pulses per second and
pitch refers to the way we perceive it.
Hi...
Miettas script...
http://www.helsinki.fi/~lennes/praat-scripts/public/open_all_files_in_folder.pra\
at
form Open all files in directory
sentence Directory /Users/YuorLogin/the_folder_youwanttoopen/
endform
Create Strings as file list... list 'directory$'*
numberOfFiles = Get number of strings
for ifile to numberOfFiles
filename$ = Get string... ifile
# You can add some filename extensions that you want to be excluded to
the next line.
if right$ (filename$, 4) <> ".doc" and right$ (filename$, 4) <> ".xls"
and right$ (filename$, 4) <> ".XLS" and right$ (filename$, 4) <> ".TXT" and
right$ (filename$, 4) <> ".txt" and right$ (filename$, 4) <> ".dat" and right$
(filename$, 4) <> ".DAT"
Read from file... 'directory$''filename$'
endif
select Strings list
endfor
select Strings list
Remove
...only changed example dir... works fine...
J
> To: praat-users@...
> From: ycsgeorge@...
> Date: Tue, 17 Nov 2009 02:13:51 +0000
> Subject: [praat-users] Praat script to open all files in a folder for Mac
>
> Dear all,
>
> I have just switched from PC to Mac and have not become so familiar with Mac.
I am wondering whether there is a particular script to open all fiels in a
folder in Mac. I tried the script that works in PC, but it does not work on my
Mac. Your script or suggestion is greatly appreciated.
>
> Best,
>
> Chunsheng
>
>
>
> ------------------------------------
>
> To Post a message, send it to: praat-users@eGroups.com
>
> To Unsubscribe, send a blank message to: praat-users-unsubscribe@eGroups.com
>
> To consult archives : http://egroups.com/list/praat-users/
> Yahoo! Groups Links
>
>
>
_________________________________________________________________
Hitta hetaste singlarna på MSN Dejting!
http://dejting.se.msn.com/channel/index.aspx?trackingid=1002952
[Non-text portions of this message have been removed]
Hi..
I suppose it depends on what you want to port it to?
For several of the purposes I use it for this will do
To MFCC... nr_coeff window_dur time_step mel_filter filter_distance max_freq
To Matrix
Transpose
Write to headerless spreadsheet file... /Users/xname/untitled.txt
Best regards
Jonas
> To: praat-users@...
> From: spchss1@...
> Date: Thu, 19 Nov 2009 06:32:01 -0800
> Subject: [praat-users] copying matrix into file
>
> hi
> I have calculated MFCC using pratt.How do I save the matrix obtained into a
file so that can be distributed to other system. Need help being a new user.
> by
>
>
> The INTERNET now has a personality. YOURS! See your Yahoo! Homepage.
http://in.yahoo.com/
>
> [Non-text portions of this message have been removed]
>
>
>
> ------------------------------------
>
> To Post a message, send it to: praat-users@eGroups.com
>
> To Unsubscribe, send a blank message to: praat-users-unsubscribe@eGroups.com
>
> To consult archives : http://egroups.com/list/praat-users/
> Yahoo! Groups Links
>
>
>
_________________________________________________________________
Hitta hetaste singlarna på MSN Dejting!
http://dejting.se.msn.com/channel/index.aspx?trackingid=1002952
[Non-text portions of this message have been removed]
At 19:43 +0000 19/11/09, hncsbei wrote:
>when I used mingw 5.1.6 to compile praat sources on windows xp, the
>error message was following:
>
>D:\MinGW\bin\GSL>mingw32-make
>touch libgsl.a
>process_begin: CreateProcess(NULL, touch libgsl.a, ...) failed.
touch, rm and ar are Unix commands, not necessarily available on
Windows. The makefiles were designed for building Praat on Unix. For
Windows we don't use MinGW, so if you want to use MinGW you will have
to do some editing of the makefiles. It will help if you have some
experience in developing applications for Windows.
--
Paul Boersma
Phonetic Sciences, University of Amsterdam
Spuistraat 210, room 303
1012VT Amsterdam, The Netherlands
http://www.fon.hum.uva.nl/paul/
phone +31-20-5252385
Dear Paul and everyone,
when I used mingw 5.1.6 to compile praat sources on windows xp, the error
message was following:
D:\MinGW\bin\GSL>mingw32-make
touch libgsl.a
process_begin: CreateProcess(NULL, touch libgsl.a, ...) failed.
make (e=2): ϵͳÕÒ²»µ½Ö¸¶¨µÄÎļþ¡£(system couldnot find the designated file)
mingw32-make: *** [libgsl.a] Error 2
What can I do for this problem?
Thank you.
hi
I have calculated MFCC using pratt.How do I save the matrix obtained into a file
so that can be distributed to other system. Need help being a new user.
by
The INTERNET now has a personality. YOURS! See your Yahoo! Homepage.
http://in.yahoo.com/
[Non-text portions of this message have been removed]
Hi everybody,
In scripts the function Get spectral power at cursor cross is available as well
as Get frequency at frequency cursor.
I had no problem to set the horizontal (Time line) cursor by "Move cursor to...
" but how can set the frequency cursor?
Thanks..
Eddy
On 2009 Nov 19, at 06:31, Jenny wrote:
> In the change gender function, if we set a new pitch median, the new
> pitch will be calculated as: oldPitch*(newPitchMedian/
> oldPitchMedian). Then not all pitch values get a same change.
>
> I am wondering why the new pitch is calculated in this way?
>
> Why not oldPitch + (newPitchMedian - oldPitchMedian). In this way,
> pitch contour moves up or down, and also keep the same shape.
>
It's all a question of scale.
What you say is true if we are sure that the appropriate scale for
pitch is the linear measurement of F0 (ie Hertz).
There are many other scales we could use. Suppose for example we
decided that what is relevant is not the value in Hz buttthe duration
of the pitch period. This is the reciprocal of the frequency (ie p = 1/
Hz) , and in one sense is a much more direct physical measurement than
the frequency itself. If we wanted to be sure that we kept the same
shape for a curve showing the pitch period instead of the frequency we
would need to do:
1/newPitch = 1/oldPitch + (1/newPitchMedian - 1/oldPitchMedian))
Phoneticians and musicians have long established that what we perceive
as pitch is more closely related to the logarithm of the frequency (log
(Hz)) than to the frequency itself. This is called the Weber-Fechner
law which states that the perception of the intensity of a stimulus
is proportional to the logarithm of the magnitude of the stimulus.
(http://en.wikipedia.org/wiki/Weber%E2%80%93Fechner_law
).
Musical instruments, for example, are tuned so that "the same number
of semitones" on the instrument corresponds to the same frequency
interval, ie the same value of log(Hz).
So to "keep the same shape" *perceptually*, we need to make changes on
a log scale, like this:
log(newPitch) = log(oldPitch) + (newPitchMedian) - log(oldPitchMedian))
Now, since we know that log (a x b) = log (a) + log (b) and
that log (a / b) = log (a) - log (b)
it follows that:
log(newPitch) = log(oldPitch * (newPitchMedian / oldPitchMedian))
and so that:
newPitch = oldPitch * (newPitchMedian / oldPitchMedian)
Hope this helps!
daniel
--------------------------------------------------------------------------------\
------------------
CNRS Laboratoire Parole et Langage (UMR 6057) - salle B004
Université de Provence, 5 avenue Pasteur
BP 80975
13604 Aix-en-Provence, Cedex 1, France
tel: +33-4-4295-3628
secr: +33-4-4295-3634
--------------------------------------------------------------------------------\
-------------------
Speech Prosody V - May 2010 Chicago, Illinois, USA
http://www.speechprosody2010.illinois.edu
Speech Prosody list: http://uk.groups.yahoo.com/group/speech_prosody
Momel-Intsint list: http://tech.groups.yahoo.com/group/momel-intsint
Praat-users list:
http://groups.yahoo.com/group/praat-users
List of mypublications:
http://www.lpl.univ-aix.fr/publi/57
--------------------------------------------------------------------------------\
-------------------
[Non-text portions of this message have been removed]
Dear All,
In the change gender function, if we set a new pitch median, the new pitch will
be calculated as: oldPitch*(newPitchMedian/oldPitchMedian). Then not all pitch
values get a same change.
I am wondering why the new pitch is calculated in this way?
Why not oldPitch + (newPitchMedian - oldPitchMedian). In this way, pitch contour
moves up or down, and also keep the same shape.
Any help will be appreciated.
Thanks in advance.
Jenny
Hello all. I am trying to combine a KlattGrid object with a sound to use vocal
tract filtering. I can do this successfully with the cascade filter model.
However, when I try to use the parallel filter model, Praat crashes. I have the
latest version of Praat (5.1.20) and am running Windows XP. Any help would be
greatly appreciated.
Best, Kyna Betancourt
Dear all,
I have just switched from PC to Mac and have not become so familiar with Mac. I
am wondering whether there is a particular script to open all fiels in a folder
in Mac. I tried the script that works in PC, but it does not work on my Mac.
Your script or suggestion is greatly appreciated.
Best,
Chunsheng
At 21:39 +0800 16/11/09, Li Yunjing wrote:
> Praat actually provides two ways of making a spectrum. One is to
>move cursor to a point for example 0.122s in Sound editor, then
>click "View spectral slice", then a Spectrum object name xx_0.122
>will appear in the Praat object window.
yes, this makes a complex-valued spectrum (in Pa/Hz) from the
selected part of the sound (or from a part that is one analysis
window long, if you only select one time point). This spectrum
contains all the information needed to restore the orignal windowed
sound.
> The other way is to select the Sound object in the Praat object
>window, then click "To spectrogram", then a Spectrogram object will
>appear in the Praat object window. Then click "To spectrum (slice)",
>(the point to extract the spectrum is also 0.122), a Spectrum object
>will appear in the Praat object window.
When Praat computes a spectrogram, the complex-value information (and
thus the timing of the components) is lost, because the spectrogram
is just a power spectral density (in Pa-squared/Hz). The Spectrum is
computed as its real-valued square root, basically; this means you
cannot restore the original windowed sound from it by doing "To
Sound". Also, the frequency resolution of a spectrogram is typically
much coarser than that of a Spectrum; otherwise it wouldn't easily
fit in memory.
--
Paul Boersma
Phonetic Sciences, University of Amsterdam
Spuistraat 210, room 303
1012VT Amsterdam, The Netherlands
http://www.fon.hum.uva.nl/paul/
phone +31-20-5252385
I've been logging vowel formant contours, so far by modeling them as
complex polynomials (up to 9 coefficients). It's straightforward with
Praat's inbuilt linear regression function. But I'm trying to switch
to smoothing spline functions, to avoid some of the unwanted
properties of polynomials. I understand that Praat can create and work
with splines, but only if the coefficients are already known. What I
need is a way to calculate spline coefficients from a series of data
points in the first place.
Unfortunately, the math involved in spline functions is a bit over my
head still. Does anyone have code for this? Or any suggestions on how
to implement smoothing spline regression in Praat?
Thanks!
--Chris
Hi all,
I met a problem in making spectrums.
Praat actually provides two ways of making a spectrum. One is to move
cursor to a point for example 0.122s in Sound editor, then click Spectrum¨¤View
spectral slice, then a Spectrum object name xx_0.122 will appear in the Praat
object window, like the picture below.
The other way is to select the Sound object in the Praat object window,
then click Spectrum¨¤To spectrogram, then a Spectrogram object will appear in
the Praat object window. Then click Analyse¨¤To spectrum (slice), (the point to
extract the spectrum is also 0.122), a Spectrum object will appear in the Praat
object window, like the picture below.
These two ways are the same in theory, right? Yes, these two pictures are
similar, but two differences exist:
1) the dB values on the right edge are different. In the first picture those are
34.6 and -25.4 dB, but in the second picture they are 102.0 and 42.0 dB.
2) the contours in the two pictures are a little different. The formants
observed from these two pictures maybe different too.
Can anyone explain this for me? Which one of these two results is correct?
Thank you very much in advance!
Li Yunjing
[Non-text portions of this message have been removed]
Has anyone had any experience in trying to use right-to-left text in
instructions in ExperimentMFC, for e.g. Arabic, Farsi or other languages?
For a current experiment I would like to provide my listeners with start text
etc instructions in Arabic. I can get the individual words to appear correctly
but the order of the words is reversed (appearing in left-to-right rather than
right-to-left order). Any suggestions?
Many thanks,
Sam
Hello,
I'm working on conversational corpora, and wondering what sorts of preparation
of the audio files would be ideal prior to transcription? What filters should I
apply, and what other settings in Praat are useful to adjust?
Also, do you know of any available Praat transcription tools I should be using?
I was planning to build a set, but don't want to recreate what's already around.
Thanks,
db
At 10:23 +0000 12/11/09, rachel.baker1981 wrote:
>However, I don't want to just exclude 'and', I want to exclude all
>intervals containing any function word. I looked on the forum for
>suggestions about how to add more than one condition to the IF
>command and tried the two below:
>
>#if intervalTextOld$ <> ":and:for:it:"
>#if intervalTextOld$ <> "and" or intervalTextOld$ <> "for" or
>intervalTextOld$ <> "it"
>
>Neither of these worked.
The second one works. Use "..." if you want to distribute the
condition over multiple lines.
> Because I will have a long list of function words that need to be
>excluded, e.g. and, for, it, on, to etc...would it make more sense
>to create a list file containing all the 'exclusion' words, program
>the script to loop through this file and only carry out the copying
>if the interval does not match any of the words in the list?
Yes. You can use "Read Strings from raw text file...", then cycle
through the strings. Or put them into a Table file (each word on its
own line, but a header on the first line), then use "Read Table from
tab-separated file...", then use "Search column...".
--
Paul Boersma
Phonetic Sciences, University of Amsterdam
Spuistraat 210, room 303
1012VT Amsterdam, The Netherlands
http://www.fon.hum.uva.nl/paul/
phone +31-20-5252385
Can anyone help me calculate the group delay of an auditory filter at its center
frequency?
I am using the usual rounded exponential model where filter weight is equal to
(1+pg)e**(-pg). p is the passband parameter and g is the frequency deviation
from center frequency divided by center frequency (the filter is symmetric).
Thanks, Ken Henry
Hi i am totally new to Praat. Been reading the manual and I am interested to
know more about the function of the formant filters.
when do i need to add in a formant filter and how do i use it?
Lastly, does anyone have a list of the commands for writing Klattgrid scripts?
Thanks =)
mizu
I am trying to write a script which duplicates intervals on TextGrid tier 1 to
TextGrid tier 2 only if the interval text does not contain a function word. So
far, I have got a script running which will copy all intervals except if they
contain the text "and"
# If non-function word, copy interval to tier 2
if intervalTextOld$ <> "and"
# Get the interval boundaries of the mixed tier
startTime = Get starting point... 1 iIntervalOld
endTime = Get end point... 1 iIntervalOld
ETC....
However, I don't want to just exclude 'and', I want to exclude all intervals
containing any function word. I looked on the forum for suggestions about how to
add more than one condition to the IF command and tried the two below:
#if intervalTextOld$ <> ":and:for:it:"
#if intervalTextOld$ <> "and" or intervalTextOld$ <> "for" or intervalTextOld$
<> "it"
Neither of these worked. Because I will have a long list of function words that
need to be excluded, e.g. and, for, it, on, to etc...would it make more sense to
create a list file containing all the 'exclusion' words, program the script to
loop through this file and only carry out the copying if the interval does not
match any of the words in the list?
Any suggestions would be much appreciated!
Rachel
i'm trying to synthesize nasals but to no avail...
Could someone kindly guide me along?
Also, are there any Praat scripts available online for consonants?
-mizu
At 16:47 +0200 4/11/09, Nele Salveste wrote:
> My computer works on Windows and I'm really wondering, why can't I
>insert pictures into my Picture window! Why are the Macintosh
>computers favored in this case?
The Mac has a function (in Quartz) that can read a JPEG or TIFF file
into an image (pixel map) in memory, and I know how to draw those to
the screen. If you supply me with a comparable function for Windows,
I can introduce picture insertion for Windows as well.
--
Paul Boersma
Phonetic Sciences, University of Amsterdam
Spuistraat 210, room 303
1012VT Amsterdam, The Netherlands
http://www.fon.hum.uva.nl/paul/
phone +31-20-5252385
At 14:51 +0000 8/11/09, Zoltan wrote:
> (Let us suppose that the fricative interval begins at "1 ms" and
>ends at "2 ms".)
>3. On harmonicity object: Get mean... 1 2 [start and end of fricative]
this should be instead: Get mean... 0.001 0.002
>I tested the methods on some files, and they consistently show
>different values for the mean (often A had "undefined" values, B
>never did).
It should be the reverse: method B will often give "undefined"
results, because the sound is often too short to contain even one
analysis frame.
The best way is indeed to cut out the fricative, so that you can be
sure that no parts of the surrounding vowels are included in the
analysis. However, since the HNR analysis requires a
non-zero-duration analysis frame, which should be able to capture at
least six periods of vocal fold vibration (for [z]), the part of the
fricative near the centre will be waited more strongly than at the
end.
--
Paul Boersma
Phonetic Sciences, University of Amsterdam
Spuistraat 210, room 303
1012VT Amsterdam, The Netherlands
http://www.fon.hum.uva.nl/paul/
phone +31-20-5252385
Suppose I want to measure the mean HNR (or to be precise, the
periodicity-to-noise ratio) of fricatives (say, /s/ vs. /z/). I thought of two
ways of doing this; however, each case produces different results. (Let us
suppose that the fricative interval begins at "1 ms" and ends at "2 ms".)
Method A:
1. Sound and TextGrid objects are loaded (TextGrid has intervals, including the
fricative portion)
2. Create Harmonicity object (To Harmonicity (cc)... 0.01 75 0.1 1)
3. On harmonicity object: Get mean... 1 2 [start and end of fricative]
Method B:
1. Sound and TextGrid objects are loaded (TextGrid has intervals, including the
fricative portion)
2. Cut out the fricative portion (Extract intervals where...) - this creates a
sound object of the fricative portion only (between 1 and 2 ms)
3. Create Harmonicity object on this new sound object (To Harmonicity (cc)...
0.01 75 0.1 1)
4. On this harmonicity object: Get mean... 0 0 (the whole portion of the
fricative, which is obviously 1 ms long)
I tested the methods on some files, and they consistently show different values
for the mean (often A had "undefined" values, B never did). To my mind, B's
values are plausible. Can you tell me which method is preferred (and why)? I
guess I am missing something obvious, but I can't really see what... Thanks.
Offline 
Send Email
