Besides doing what our colleagues have already suggested in this list, you could
configure your Windows to always show file extensions. This way you won't have
to type ".wav" when saving your wav files.

To do that, open Windows Explorer, click on the "Organize" menu, Folder and
search options, View, and, in Advanced settings, make sure the "Hide extensions
for known file types" option is not checked.

All the best,

Andreia

>
>
> I'm sorry if this has been posted before. I cannot find the answer in
> the forum. I am trying to save a selected part of my recordings as .wav
> files but for some reason it only gives me ".file" which can only be
> used in praat. How do I get it to give me a .wav file? I have noticed
> that at times I do get a .wav file but I don't know what I'm doing
> different in order for that to happen.
>

At least you have given me a logical explanation. ;-)
Thanks a lot.

*Lilla*

--- In praat-users@..., José Joaquín Atria <jjatria@...> wrote:
>
> Praat's formant detection has worked fine for me in the past as it is,
> but you have to be careful when changing the values in the formant
> settings. You can't simply change the amount of formants to look for
> and leave the "Maximum formant" field as it is.
>
> If you do, your values _will_ be off because you are basically telling
> Praat to look for too many (or too few) formants in the same frequency
> space. The default value of 5500 Hz for the maximum formant is good
> for five formants (no more, no less). If you need less (or more)
> formants, you'll need to change that value as well. However, I've had
> trouble finding a different maximum formant value that works as well
> as that for a different amount of formants, so normally I stick to the
> default values even if I need less formants for my own measurements.
>
> Hope it helps.
>
> José Joaquín Atria
>
> www.pinguinorodriguez.cl
>
>
>
> 2012/1/31 alephoncristia <alecristia@...>:
> > Hi Lilla,
> >
> > Have you looked at Escudero et al. 2009 JASA paper? They convincingly argue
that it's difficult to set parameters once and for all (talkers and vowels), and
that perhaps different parameters should be used in each case. They propose a
data-driven method, where you determine which parameter setting leads to the
most stable measurements by measuring F1 and F2 in many vowels within each
talker, and then calculating variance within each parameter setting.
> >
> > I've implemented an algorithm to do the measurements in praat:
> >
https://sites.google.com/site/acrsta/nonresearch/simple-vowels.txt?attredirects=\
0
> > and the selection in R:
> >
https://sites.google.com/site/acrsta/nonresearch/selectModel.R?attredirects=0
> >
> > Good luck!
> >
> > -alex
> > --- In praat-users@..., José Joaquín Atria <jjatria@>
wrote:
> >>
> >> Are you getting the values from a Formant object you created, or just from
> >> the editor window?
> >> What parameters are you using?
> >> What formant values are you getting? What do you think you _should_ be
> >> getting?
> >>
> >> That would be a good start.
> >>
> >> José Joaquín Atria
> >>
> >> www.pinguinorodriguez.cl
> >>
> >>
> >>
> >> On Tue, Jan 31, 2012 at 1:43 PM, latinista90 <latinista90@> wrote:
> >>
> >> > **
> >> >
> >> >
> >> > Hi!
> >> >
> >> > I'm analiysing a corpus of spontaneous italian speech and I'm under the
> >> > impression that the values of the first and the second formant of
marginal
> >> > vowel (i, u) are always too low. I suppose I must maybe change something
in
> >> > the formant settings but I don't know what..Is there anybody that can
help
> >> > me?
> >> > Thanks a million.
> >> >
> >> > *Lilla*
> >> >
> >> >
> >> >
> >>
> >>
> >> [Non-text portions of this message have been removed]
> >>
> >
> >
>

As Paul Boersma pointed out, there *are* several different interpretations and
formulae of the mel scale around.

Where does the one used in praat come from? I could not find any literature
using that formula. How has it been derived?

Thanks a lot
Anna

I have a .wav file containing a word (let's say, "orange"). What is the
easiest way to find out whether the pitch of that .wav file looks more like
a rising edge or falling edge? In other words, I would like to characterize
the pitch of that .wav file as rising edge, falling edge, or flat. How do I
do that?


[Non-text portions of this message have been removed]

Hi profs,
I am trying to isolate the last syllable from the whole utterance in Praat
(actually they are Chinese utterances and the last syllable equals the last
character). While I can do this by hand, I am wondering whether any of you has a
script by running which I can do this as a batch, or any other shortcuts (cause
I have 600-ish tokens, and that will take quite a while if I cut it one by one).
Please let me know your ideas, any suggestion is welcome and appreciated! Many
thanks.

Hi 
you need to find the time of the maximum pitch, and time of minimum pitch and
subtract ( tmax-tmin). if the product is positive then the contour is rising,
otherwise it is falling.
to get max pitch ctrl+shift+H, and then you get the time on the vertical line,
to get min pitch ctrl+shift+L
best
ayman

[Non-text portions of this message have been removed]

Hi,
re history of the mel scale, check the discussion on the Auditory list
http://www.auditory.org/ and especially Donald Greenwood's message on 23
July 2009 (http://www.auditory.org/postings/2009/581.html).

Best wishes, - Jyrki

On 03/02/2012 12:55, anna.janska wrote:
>
> As Paul Boersma pointed out, there *are* several different
> interpretations and formulae of the mel scale around.
>
> Where does the one used in praat come from? I could not find any
> literature using that formula. How has it been derived?
>
> Thanks a lot
> Anna
>
>



[Non-text portions of this message have been removed]

Dear Jyrki,

thank you for the reference; although I had already been familiar with the
elaborate discussion on the auditory list, I had not yet given particular
consideration to this very interesting post you referenced.

I am wondering, why the specific values used in Praat were used; so in other
words: the main discussion about the mel scale I have come across so far debates
why log(1+f/700) or log(1+f/1000) should be used. Praat uses log(1+f/550),
resulting in a more shallow curve. Thus I was hoping somebody could point me to
a specific reference on why that very value is used, rather than any of the
others?

Thank you,
Anna

--- In praat-users@..., Jyrki Tuomainen <jtuomain@...> wrote:
>
> Hi,
> re history of the mel scale, check the discussion on the Auditory list
> http://www.auditory.org/ and especially Donald Greenwood's message on 23
> July 2009 (http://www.auditory.org/postings/2009/581.html).
>
> Best wishes, - Jyrki
>
> On 03/02/2012 12:55, anna.janska wrote:
> >
> > As Paul Boersma pointed out, there *are* several different
> > interpretations and formulae of the mel scale around.
> >
> > Where does the one used in praat come from? I could not find any
> > literature using that formula. How has it been derived?
> >
> > Thanks a lot
> > Anna
> >
> >
>
>
>
> [Non-text portions of this message have been removed]
>

Hi,

I am attempting to write a pratt script that takes an input other than the
microphone.  I'm using Soundflower: http://code.google.com/p/soundflower/

The line I'm trying is:

Record Sound (fixed time)... "Soundflower (2ch)" 0.99 0.5 44100 recordTime

I'm new to Praat scripting, so any help is appreciated.

Best Regards,
Rob

Does anyone know if it is possible to open a Sound with the belonging textgrid
and formantgrid simultaneously?

Or only formant +textgrid?

Thanks in advance,
Anne

Hi all,

I would like to put the picture of wave form of sound in Praat picture along
with pitch and tiers. Is is possible in Praat?


Best regards
Susanto
www.mrsusanto.co.cc

Praat can't do it automatically out of the box, but you can draw them one
by one making sure they fit your layout and have the information you need.

Once you are content with your results, you can put all those instructions
into a script and run that script every time you wish to draw the same
combination of objects.

You'd end up with something like this (but hopefully a lot more polished):

> # Draw Sound in black
> Black
> select yoursound
> Select outer viewport... 0 5.5 0 3
> Draw... 0 0 0 0 no Curve
> # Draw Pitch in blue
> select yourpitch
> Blue
> Draw... 0 0 0 500 no
> # Draw TextGrid in black
> # Note that the viewport is larger so that tiers fit
> # How much larger will depend on the amount of tiers it has
> Select outer viewport... 0 5.5 0 5
> Black
> select yourtextgrid
> Draw... 0 0 yes yes yes

If you absolutely can't manage to get your perfect figure, you can always
save the drawing as it is and modify it later using a vector graphics
program, like Inkscape, which is free. But if you have to make a lot of
them then this extra step might not be as helpful.

José Joaquín Atria

www.pinguinorodriguez.cl



On Wed, Feb 8, 2012 at 1:18 AM, susanto <susantonanda@...> wrote:

> **
>
>
> Hi all,
>
> I would like to put the picture of wave form of sound in Praat picture
> along with pitch and tiers. Is is possible in Praat?
>
> Best regards
> Susanto
> www.mrsusanto.co.cc
>
>
>


[Non-text portions of this message have been removed]

Hello everyone
So my story is that I'm trying to create a rating program so that participants
can rate audio files on the basis of specific characteristics of speech. I'm
currently getting multiple errors and would really appreciate your help with
these errors. My first error is "found a number while looking for a string in
text (line 77), second is string "name" not read, third is Experiment MFC not
read, fourth is collection not read, fifth is data not read from text file,
sixth is no object was put into the list. I'm really confused about these errors
and any help would be greatly appreciated. I can uploade the script file if this
could be useful.

Op 8 feb 2012, om 21:22 heeft louisdube28 het volgende geschreven:

> > I'm currently getting multiple errors and would really appreciate your help
with these errors. My first error is "found a number while looking for a string
in text (line 77), second is string "name" not read, third is Experiment MFC not
read, fourth is collection not read, fifth is data not read from text file,
sixth is no object was put into the list.

this is all the same error. You forgot to type one of the elements of the
ExperimentMFC, or forgot a quote, or typed a number too many. See the tutorial
for examples. Nothing in them can be skipped.

If this doesn't help, copy the ExperimentMFC file in your next message.
_____

Paul Boersma
University of Amsterdam
Linguistics
Spuistraat 210, room 303
1012VT Amsterdam, The Netherlands
http://www.fon.hum.uva.nl/paul/

Hello all,

I'm not exactly sure when this starting happening, but whenever I try to record
mono sound a window pops up that says the error described in the subject has
occurred. I am then forced to force quit the application. I tried updating to
the newest version of Praat but no luck. Has anybody else had a similar problem?
When I searched for the error here and in google I got back no hits. Thanks in
advance.

I am new to Praat and could use some guidance.  I would like to write a script
that locates the local maxima of a waveform (a plucked stringed instrument) and
marks it with an interval tier.  I have been reading manuals but am no closer to
figuring this problem out.

Thank you.

On 15 Feb 2012, at 20:58 , praatquestion wrote:

> I am new to Praat and could use some guidance. I would like to write a script
that locates the local maxima of a waveform (a plucked stringed instrument) and
marks it with an interval tier. I have been reading manuals but am no closer to
figuring this problem out.

You could do that with a script like the one at the bottom of this message. Copy
the script, choose New Praat script from the Praat menu,  paste the script into
it (and save it as get_peaks.praat), select a Sound and choose Run from the Run
menu of the script.

The script should display the Sound and a TextGrid with one interval beginning
at each peak. The script works through the values of intensity from right to
left so that the peaks found are at the beginning rather than the end of a
plateau.

The script assumes that a peak is above a minimum value of dB (Min_dB) and is
more than 'Threshold' above neighbouring valleys.
You need to specify the minimum pitch you expect to find. The default here is 70
Hz (just over C#2).

You will need to experiment with different values of Min_dB and Threshold - the
default values (50 and 1) worked fairly well for a short extract of a recording
of a guitar.
When the TextGrid opens, choose Show Intensity from the Intensity menu. You
would probably do well to switch off the other displays (ie Pitch, Formant,
Spectrum, Pulses).
You can then try to find optimum values for the two parameters by looking at the
relative intensity values of what you think are errors (either false positives
or false negatives).

This is just a quick and dirty script to show you how you can tackle the
problem. You might improve it by adding things like a minimum duration between
two peaks etc.

best

daniel



> # Praat script: get_peaks.praat
> # Author: Daniel Hirst <daniel.hirst@...>
> # Version: [2012-02-16]
>
> # Purpose: Create TextGrid with point tier for intensity peaks
> # Requires: One Sound (selected)
>
> # Note: Peaks must be above Min_dB and more than Threshold above neighbouring
valleys.
>
> form Get peaks
>  positive Min_pitch 70 (=Hz)
>  positive Time_step 0.01 (= s)
>  positive Min_dB 50 (=dB)
>  positive Threshold 1 (= dB)
> endform
>
>
> mySound = selected("Sound")
> myTextGrid = To TextGrid... Peaks
>
> select mySound
> myIntensity = To Intensity... min_pitch time_step yes
>
> nPoints = Get number of frames
>
> iPoint = nPoints
> right_value = 0
> this_value = Get value in frame... iPoint
>
> while iPoint > 0
>  select myIntensity
>  if iPoint = 1
> 	 left_value = 0
>  else
> 	 left_value = Get value in frame... iPoint-1
>  endif
>
>  if (this_value-right_value)>threshold and (this_value-left_value)>threshold
>  ... and this_value > min_dB
> ## It's a peak
> 	 right_value = this_value
> 	 time = Get time from frame number... iPoint
> 	 select myTextGrid
> 	 Insert boundary... 1 time
>  elsif (right_value-this_value) > threshold and
(left_value-this_value)>threshold
> ## It's a valley so remember it
> 	 right_value = this_value
>  endif
>  this_value = left_value
>  iPoint -= 1
> endwhile
>
> select myIntensity
> Remove
>
> select mySound
> plus myTextGrid
> View & Edit
> select mySound


--------------------------------------------------------------------------------\
------------------
alternative email:                 mailto:daniel.hirst@...

        CNRS Laboratoire Parole et Langage (UMR 7309) - salle B012
        Université de Provence, 5 avenue Pasteur  BP 80975
             ⬇    PLEASE NOTE NEW TELEPHONE NUMBERS     ⬇
  tel: +33-4-1355-3628                                             secr:
+33-4-1355-3632/4
                             13604 Aix-en-Provence, Cedex 1, France
--------------------------------------------------------------------------------\
-------------------
Speech Prosody SIG 	 http://aune.lpl.univ-aix.fr/projects/sprosig/
Speech Prosody list:   http://uk.groups.yahoo.com/group/speech_prosody
Praat-users list:                          
http://groups.yahoo.com/group/praat-users
List of my publications:                                
http://www.lpl-aix.fr/publi/57
--------------------------------------------------------------------------------\
-------------------

Hi,

I am a science honours student at the University of Queensland and I am looking
for people who use Praat in Brisbane, Australia.

Thank you very much for your time

Cassandra


[Non-text portions of this message have been removed]

Dear All,
I'm trying to speech analysis, and need to use KlattGrid values to synthesis
vowels. So I tried to write a praat script.

select Sound 'base_name$'

To KlattGrid(simple)... 0.005 5 5500 0.025 50.0 60.0 600.0 100.0 yes

this part of the code give error message as 'change something on command window'
so how can i change this values, or how can i write a script for
'Get voicing amplitude at time'?
thanks all.

Op 21 feb 2012, om 13:23 heeft phclummia het volgende geschreven:

> > To KlattGrid(simple)... 0.005 5 5500 0.025 50.0 60.0 600.0 100.0 yes
>
> this part of the code give error message as 'change something on command
window'

a more informative part of the error message probably came before that. One
thing I see is that you forgot to insert a space before the first parenthesis.
_____

Paul Boersma
University of Amsterdam
Linguistics
Spuistraat 210, room 303
1012VT Amsterdam, The Netherlands
http://www.fon.hum.uva.nl/paul/

Hej José,
thanks for your help. There is one error I get with your script:

"Interval number" must be a positive whole number.
Don't understand contents of field "Interval number". Script line 65 not
performed or completed:
"a = Get start point... 1 intervals-1"

Could this be due to the fact that I have long stretches of silence in my audio
files?

Thanks,
Riccardo

--- In praat-users@..., José Joaquín Atria <jjatria@...> wrote:
>
> Why don't you cut up the LongSound into smaller Sounds that you can analyze
> like you've been doing?
> I wrote a small script that does what you want, but it gets the pitch
> values differently and it stores them in a Table object instead of saving
> them to disk. Still, you should be able to modify it to fit your needs.
>
> You can find it at http://ucl.ac.uk/~ucjt465/scripts/pitch_listing.praat
>
> Hope it helps.
>
> José Joaquín Atria
>
> www.pinguinorodriguez.cl
>
>
>
> On Tue, Jan 24, 2012 at 7:27 PM, riccardo_fusaroli <fusaroli@...>wrote:
>
> > **
> >
> >
> > Hi,
> > I have been going through a few email threads on how to export pitch and
> > intensity listings to a txt file via scripting. However, most of the
> > methods suggested such as
> >
> > To Pitch... 0.012 75 600
> > Down to PitchTier
> > Down to TableOfReal... Hertz
> > Rename... pitch_'name$'
> > Write to headerless spreadsheet file... 'directory2$'\'name$'_f0.txt
> >
> > only work for files opened with Read from file...
> >
> > I am working with a corpus of long conversations, which files requires
> > Open long sound file...
> > That creates problem with the To Pitch... and other functions.
> >
> > My dirty workaround has been the following:
> >
> > Open long sound file... 'directory$'/'fileName$'
> > select LongSound 'fileName$'
> > pointS = Get start time
> > pointE = Get end time
> > View
> > editor LongSound 'fileName$'
> > Pitch settings... 75 600 Hertz autocorrelation automatic
> > Select... pointS pointE
> > length = pointE - pointS
> > Show analyses... yes yes yes no no length
> > fileName$ = fileName$ - ".wav"
> > Show all
> > pitch$ = Pitch listing
> > pitch$ > 'directory$''fileName$'_pitch.txt
> > endeditor
> >
> > However, this only works for short files. When using a file above 5
> > minutes, I only get the first 30 seconds of pitch listing.
> >
> > Thanks in advance for any help,
> > Riccardo
> >
> >
> >
>
>
> [Non-text portions of this message have been removed]
>

I'm not sure.

The only way I've been able to reproduce that error is if the To TextGrid
(silences)... detects a really long voiced interval at the beginning of
your sound (longer than the chunkLength). If that happens, then the script
will try to move the boundary of the part to the middle of the previous
interval, which of course, doesn't exist (and will die doing so since
"'interval
number' must be a positive whole number").

Maybe you could try changing the value of chunkLength, to try and avoid
having boundaries fall on voiced intervals? This is actually what I think
is the easiest way to fix this. Or if your sound definitely _does not_
start with a really long voiced interval, then maybe changing something in
the line where the VUV segments are detected would get rid of the error.
You could also uncomment the printline calls to see what's going on; maybe
that gives you a hint. Or comment out the Remove line at the end and keep
the TextGrids, so you can see what the VUV intervals were.

Let me know how that goes, and if it doesn't work, we can take it from
there.

I'll try to update the script to account for such an error.

José Joaquín Atria

www.pinguinorodriguez.cl



On Tue, Feb 21, 2012 at 1:26 PM, riccardo_fusaroli <fusaroli@...>wrote:

> **
>
>
> Hej José,
> thanks for your help. There is one error I get with your script:
>
> "Interval number" must be a positive whole number.
> Don't understand contents of field "Interval number". Script line 65 not
> performed or completed:
> "a = Get start point... 1 intervals-1"
>
> Could this be due to the fact that I have long stretches of silence in my
> audio files?
>
> Thanks,
> Riccardo
>
>
> --- In praat-users@..., José Joaquín Atria <jjatria@...>
> wrote:
> >
> > Why don't you cut up the LongSound into smaller Sounds that you can
> analyze
> > like you've been doing?
> > I wrote a small script that does what you want, but it gets the pitch
> > values differently and it stores them in a Table object instead of saving
> > them to disk. Still, you should be able to modify it to fit your needs.
> >
> > You can find it at http://ucl.ac.uk/~ucjt465/scripts/pitch_listing.praat
> >
> > Hope it helps.
> >
> > José Joaquín Atria
> >
> > www.pinguinorodriguez.cl
> >
> >
> >
> > On Tue, Jan 24, 2012 at 7:27 PM, riccardo_fusaroli <fusaroli@...>wrote:
> >
> > > **
>
> > >
> > >
> > > Hi,
> > > I have been going through a few email threads on how to export pitch
> and
> > > intensity listings to a txt file via scripting. However, most of the
> > > methods suggested such as
> > >
> > > To Pitch... 0.012 75 600
> > > Down to PitchTier
> > > Down to TableOfReal... Hertz
> > > Rename... pitch_'name$'
> > > Write to headerless spreadsheet file... 'directory2$'\'name$'_f0.txt
> > >
> > > only work for files opened with Read from file...
> > >
> > > I am working with a corpus of long conversations, which files requires
> > > Open long sound file...
> > > That creates problem with the To Pitch... and other functions.
> > >
> > > My dirty workaround has been the following:
> > >
> > > Open long sound file... 'directory$'/'fileName$'
> > > select LongSound 'fileName$'
> > > pointS = Get start time
> > > pointE = Get end time
> > > View
> > > editor LongSound 'fileName$'
> > > Pitch settings... 75 600 Hertz autocorrelation automatic
> > > Select... pointS pointE
> > > length = pointE - pointS
> > > Show analyses... yes yes yes no no length
> > > fileName$ = fileName$ - ".wav"
> > > Show all
> > > pitch$ = Pitch listing
> > > pitch$ > 'directory$''fileName$'_pitch.txt
> > > endeditor
> > >
> > > However, this only works for short files. When using a file above 5
> > > minutes, I only get the first 30 seconds of pitch listing.
> > >
> > > Thanks in advance for any help,
> > > Riccardo
> > >
> > >
> > >
> >
> >
> > [Non-text portions of this message have been removed]
> >
>
>
>


[Non-text portions of this message have been removed]

The new script should at least account for the error. It now also offers
the possibility to ignore the VUV check, but that might introduce some
errors in the pitch values.

Let me know if it helped.

José Joaquín Atria

www.pinguinorodriguez.cl



2012/2/21 José Joaquín Atria <jjatria@...>

> I'm not sure.
>
> The only way I've been able to reproduce that error is if the To TextGrid
> (silences)... detects a really long voiced interval at the beginning of
> your sound (longer than the chunkLength). If that happens, then the script
> will try to move the boundary of the part to the middle of the previous
> interval, which of course, doesn't exist (and will die doing so since
"'interval
> number' must be a positive whole number").
>
> Maybe you could try changing the value of chunkLength, to try and avoid
> having boundaries fall on voiced intervals? This is actually what I think
> is the easiest way to fix this. Or if your sound definitely _does not_
> start with a really long voiced interval, then maybe changing something in
> the line where the VUV segments are detected would get rid of the error.
> You could also uncomment the printline calls to see what's going on; maybe
> that gives you a hint. Or comment out the Remove line at the end and keep
> the TextGrids, so you can see what the VUV intervals were.
>
> Let me know how that goes, and if it doesn't work, we can take it from
> there.
>
> I'll try to update the script to account for such an error.
>
> José Joaquín Atria
>
> www.pinguinorodriguez.cl
>
>
>
>
> On Tue, Feb 21, 2012 at 1:26 PM, riccardo_fusaroli <fusaroli@...>wrote:
>
>> **
>>
>>
>> Hej José,
>> thanks for your help. There is one error I get with your script:
>>
>> "Interval number" must be a positive whole number.
>> Don't understand contents of field "Interval number". Script line 65 not
>> performed or completed:
>> "a = Get start point... 1 intervals-1"
>>
>> Could this be due to the fact that I have long stretches of silence in my
>> audio files?
>>
>> Thanks,
>> Riccardo
>>
>>
>> --- In praat-users@..., José Joaquín Atria <jjatria@...>
>> wrote:
>> >
>> > Why don't you cut up the LongSound into smaller Sounds that you can
>> analyze
>> > like you've been doing?
>> > I wrote a small script that does what you want, but it gets the pitch
>> > values differently and it stores them in a Table object instead of
>> saving
>> > them to disk. Still, you should be able to modify it to fit your needs.
>> >
>> > You can find it at
>> http://ucl.ac.uk/~ucjt465/scripts/pitch_listing.praat
>> >
>> > Hope it helps.
>> >
>> > José Joaquín Atria
>> >
>> > www.pinguinorodriguez.cl
>> >
>> >
>> >
>> > On Tue, Jan 24, 2012 at 7:27 PM, riccardo_fusaroli <fusaroli@...>wrote:
>> >
>> > > **
>>
>> > >
>> > >
>> > > Hi,
>> > > I have been going through a few email threads on how to export pitch
>> and
>> > > intensity listings to a txt file via scripting. However, most of the
>> > > methods suggested such as
>> > >
>> > > To Pitch... 0.012 75 600
>> > > Down to PitchTier
>> > > Down to TableOfReal... Hertz
>> > > Rename... pitch_'name$'
>> > > Write to headerless spreadsheet file... 'directory2$'\'name$'_f0.txt
>> > >
>> > > only work for files opened with Read from file...
>> > >
>> > > I am working with a corpus of long conversations, which files requires
>> > > Open long sound file...
>> > > That creates problem with the To Pitch... and other functions.
>> > >
>> > > My dirty workaround has been the following:
>> > >
>> > > Open long sound file... 'directory$'/'fileName$'
>> > > select LongSound 'fileName$'
>> > > pointS = Get start time
>> > > pointE = Get end time
>> > > View
>> > > editor LongSound 'fileName$'
>> > > Pitch settings... 75 600 Hertz autocorrelation automatic
>> > > Select... pointS pointE
>> > > length = pointE - pointS
>> > > Show analyses... yes yes yes no no length
>> > > fileName$ = fileName$ - ".wav"
>> > > Show all
>> > > pitch$ = Pitch listing
>> > > pitch$ > 'directory$''fileName$'_pitch.txt
>> > > endeditor
>> > >
>> > > However, this only works for short files. When using a file above 5
>> > > minutes, I only get the first 30 seconds of pitch listing.
>> > >
>> > > Thanks in advance for any help,
>> > > Riccardo
>> > >
>> > >
>> > >
>> >
>> >
>> > [Non-text portions of this message have been removed]
>> >
>>
>>
>>
>
>


[Non-text portions of this message have been removed]

I used the pitch listing function on Praat to measure all of the frequencies
used in a short conversation for a number of speakers. Before I measure the
measures of central tendancy, I want to trim the outliers. I have a large number
of 100-120 hz and some in the 600 or 700s even. What is considered the "normal"
speech range for adult females? Is there a published authority on this? Thank
you.

On 22 Feb 2012, at 6:14 , shannon-bailey@... wrote:

> I used the pitch listing function on Praat to measure all of the frequencies
used in a short conversation for a number of speakers. Before I measure the
measures of central tendancy, I want to trim the outliers. I have a large number
of 100-120 hz and some in the 600 or 700s even. What is considered the "normal"
speech range for adult females? Is there a published authority on this? Thank
you.

Hello Shannon - I've told you most of what follows before, but this may be
useful for others with the same problem:

A technique my students and I use is described in my paper in the first issue of
the Journal of Speech Sciences (JoSS 1(1):55-83. 2011.)
available here: 		
http://www.journalofspeechsciences.org/index.php/journalofspeechsciences/article\
/view/21

We use the median as the measure of central tendency because, unlike the mean,
it is not influenced by outliers and it is not dependent on the scale (linear or
log) used for the measurements.

For the pitch range, we use the first quartile (q1) to estimate the pitch floor
with the empirical formula:

	 floor = q1* 0.75

In earlier versions we calculated the pitch ceiling in similar fashion from the
third quartile (q3) with:

	 ceiling = q3 *1.5

however, we found that in expressive speech this value is too low - in the
article we give an example of a female speaker (from the Aix-Marsec corpus)
whose median pitch is 176 Hz and first and third quartiles 158 Hz and 208 Hz.
From this we estimate the floor and ceiling as respectively 118 Hz and 312 Hz.
On the third word of the utterance, however, the speaker's voice rises to a peak
of 433 Hz which is more than an octave above the third quartile!

In fact, the value of the pitch ceiling is less critical than that of the pitch
floor for eliminating octave errors. We consequently now use a much higher
value:

	 ceiling = q3 * 2.5

This allows us to eliminate most errors of pitch detection without manually
adjusting the values of the floor and ceiling. With a Sound selected, the
following extract of script can then be used, at least as a first estimate, for
all speakers:

>  mySound = selected("Sound")
>  To Pitch... 0.01 60 700
>  q1 = Get quantile... 0 0 0.25 Hertz
>  q3 = Get quantile... 0 0 0.75 Hertz
>  floor = q1*0.75
>  ceiling = q3*2.5
>  Remove
>  select mySound
>  To Pitch... 0.01 floor ceiling


On pitch range a few references are:

Graddol, David 1986 "Discourse specific pitch behaviour." in Catherine
Johns-Lewis (ed.) Intonation in Discourse. Croom-Helm. 221-237
Hartmut Traunmüller and Anders Eriksson (1995, manuscript) "The frequency range
of the voice fundamental in the speech of male and female adults"
(http://www2.ling.su.se/staff/hartmut/aktupub.htm)
Ladd, D.R. 1996. (2008)  Chapter 7 "Pitch range". in Intonational Phonology
pp252-283. Cambridge University Press.

Hope this is useful!

daniel




--------------------------------------------------------------------------------\
------------------
alternative email:                 mailto:daniel.hirst@...

        CNRS Laboratoire Parole et Langage (UMR 7309) - salle B012
        Université de Provence, 5 avenue Pasteur  BP 80975
             ⬇    PLEASE NOTE NEW TELEPHONE NUMBERS     ⬇
  tel: +33-4-1355-3628                                             secr:
+33-4-1355-3632/4
                             13604 Aix-en-Provence, Cedex 1, France
--------------------------------------------------------------------------------\
-------------------
Speech Prosody SIG 	 http://aune.lpl.univ-aix.fr/projects/sprosig/
Speech Prosody list:   http://uk.groups.yahoo.com/group/speech_prosody
Praat-users list:                          
http://groups.yahoo.com/group/praat-users
List of my publications:                                
http://www.lpl-aix.fr/publi/57
--------------------------------------------------------------------------------\
-------------------

I solved the problem by changing the chunklength to 600.
Thanks again!
All the best
Riccardo

2012/2/21 José Joaquín Atria <jjatria@...>

> The new script should at least account for the error. It now also offers
> the possibility to ignore the VUV check, but that might introduce some
> errors in the pitch values.
>
> Let me know if it helped.
>
> José Joaquín Atria
>
> www.pinguinorodriguez.cl
>
>
>
>
> 2012/2/21 José Joaquín Atria <jjatria@...>
>
>> I'm not sure.
>>
>> The only way I've been able to reproduce that error is if the To TextGrid
>> (silences)... detects a really long voiced interval at the beginning of
>> your sound (longer than the chunkLength). If that happens, then the script
>> will try to move the boundary of the part to the middle of the previous
>> interval, which of course, doesn't exist (and will die doing so since
"'interval
>> number' must be a positive whole number").
>>
>> Maybe you could try changing the value of chunkLength, to try and avoid
>> having boundaries fall on voiced intervals? This is actually what I think
>> is the easiest way to fix this. Or if your sound definitely _does not_
>> start with a really long voiced interval, then maybe changing something in
>> the line where the VUV segments are detected would get rid of the error.
>> You could also uncomment the printline calls to see what's going on; maybe
>> that gives you a hint. Or comment out the Remove line at the end and keep
>> the TextGrids, so you can see what the VUV intervals were.
>>
>> Let me know how that goes, and if it doesn't work, we can take it from
>> there.
>>
>> I'll try to update the script to account for such an error.
>>
>> José Joaquín Atria
>>
>> www.pinguinorodriguez.cl
>>
>>
>>
>>
>> On Tue, Feb 21, 2012 at 1:26 PM, riccardo_fusaroli <fusaroli@...>wrote:
>>
>>> **
>>>
>>>
>>> Hej José,
>>> thanks for your help. There is one error I get with your script:
>>>
>>> "Interval number" must be a positive whole number.
>>> Don't understand contents of field "Interval number". Script line 65 not
>>> performed or completed:
>>> "a = Get start point... 1 intervals-1"
>>>
>>> Could this be due to the fact that I have long stretches of silence in
>>> my audio files?
>>>
>>> Thanks,
>>> Riccardo
>>>
>>>
>>> --- In praat-users@..., José Joaquín Atria <jjatria@...>
>>> wrote:
>>> >
>>> > Why don't you cut up the LongSound into smaller Sounds that you can
>>> analyze
>>> > like you've been doing?
>>> > I wrote a small script that does what you want, but it gets the pitch
>>> > values differently and it stores them in a Table object instead of
>>> saving
>>> > them to disk. Still, you should be able to modify it to fit your needs.
>>> >
>>> > You can find it at
>>> http://ucl.ac.uk/~ucjt465/scripts/pitch_listing.praat
>>> >
>>> > Hope it helps.
>>> >
>>> > José Joaquín Atria
>>> >
>>> > www.pinguinorodriguez.cl
>>> >
>>> >
>>> >
>>> > On Tue, Jan 24, 2012 at 7:27 PM, riccardo_fusaroli <fusaroli@
>>> ...>wrote:
>>> >
>>> > > **
>>>
>>> > >
>>> > >
>>> > > Hi,
>>> > > I have been going through a few email threads on how to export pitch
>>> and
>>> > > intensity listings to a txt file via scripting. However, most of the
>>> > > methods suggested such as
>>> > >
>>> > > To Pitch... 0.012 75 600
>>> > > Down to PitchTier
>>> > > Down to TableOfReal... Hertz
>>> > > Rename... pitch_'name$'
>>> > > Write to headerless spreadsheet file... 'directory2$'\'name$'_f0.txt
>>> > >
>>> > > only work for files opened with Read from file...
>>> > >
>>> > > I am working with a corpus of long conversations, which files
>>> requires
>>> > > Open long sound file...
>>> > > That creates problem with the To Pitch... and other functions.
>>> > >
>>> > > My dirty workaround has been the following:
>>> > >
>>> > > Open long sound file... 'directory$'/'fileName$'
>>> > > select LongSound 'fileName$'
>>> > > pointS = Get start time
>>> > > pointE = Get end time
>>> > > View
>>> > > editor LongSound 'fileName$'
>>> > > Pitch settings... 75 600 Hertz autocorrelation automatic
>>> > > Select... pointS pointE
>>> > > length = pointE - pointS
>>> > > Show analyses... yes yes yes no no length
>>> > > fileName$ = fileName$ - ".wav"
>>> > > Show all
>>> > > pitch$ = Pitch listing
>>> > > pitch$ > 'directory$''fileName$'_pitch.txt
>>> > > endeditor
>>> > >
>>> > > However, this only works for short files. When using a file above 5
>>> > > minutes, I only get the first 30 seconds of pitch listing.
>>> > >
>>> > > Thanks in advance for any help,
>>> > > Riccardo
>>> > >
>>> > >
>>> > >
>>> >
>>> >
>>> > [Non-text portions of this message have been removed]
>>> >
>>>
>>>
>>>
>>
>>
>


--
--------------------
Riccardo Fusaroli, post.doctoral fellow
http://www.google.com/profiles/fusaroli
Center for Semiotics - Institute of Aesthetics and Communication
Interacting Minds - Center of Functionally Integrative Neuroscience
University of Aarhus


[Non-text portions of this message have been removed]

Hello,

I have a question concerning the ExperimentMFC tool. We would like to use it for
an experiment with quite long auditive stimuli (question-answer pairs) which the
participants have to rate on a 7-point scale.  We noticed that it is possible to
give the rating and go on to the next stimulus before the audio file has been
completely played. This is a problem for us because we do not want the
participants to skip stimuli (by accident or by purpose). Is there a solution to
this problem?


Best,

Marta Wierzba

Hi everybody!

As I try to draw vocal ellipse with the tool "formant plotter" appears this
message and I don't know what I have to do:

Column "label" does not exist.
Script line 22 not performed or completed:
« Sort rows... label »
Please change something in the window `Formant plotter', or click Cancel in that
window.

The problem is related to the file which I have to give as path, but I don't now
exactly what it does mean, that a "column does not exist".
The file that I give as path is created through the text file, which name is
"Vowelsanalyse".

Could anyone help me? Maybe have any of you experience the same problem?

Dear all,

I'm using a script that starts by defining two arrays. When running the
script a first time, everything works smoothly. However, the next time I
  run the script (on another file) Praat crashes (I'm using the latest
version, on Windows Vista). Then I tried to narrow the problem down.
When running the script a second time, even the first lines of the
definitions of the two arrays are enough to cause the crash, so it's not
  due to the length of the arrays.

I have also tried running the script a second time with array
definitions only. No problems occur in this case, so the main part of
the script seems to play some role here - I had no success in narrowing
it down any further.

Also, the problem only occurs if I close the editor window for the first
  sound/textgrid pair, and run the script on another pair.

Right now this is more of a nuisance than a problem for me, as I can
avoid it by not closing the previous editor window. Still, I can
reproduce the crash so I thought it would be worth reporting this here.

I am attaching my script, although it's quite long...

Best wishes,
Robert

---
consonants$ [1] = "N"
consonants$ [2] = "S"
consonants$ [3] = "Y"
consonants$ [4] = "Z"
consonants$ [5] = "S"
consonants$ [6] = "M"
consonants$ [7] = "TH"
consonants$ [8] = "D"
consonants$ [9] = "W"
consonants$ [10] = "K"
consonants$ [11] = "?"
consonants$ [12] = "L"
consonants$ [13] = "J"
consonants$ [14] = "H"
consonants$ [15] = "SH"
consonants$ [16] = "T"
consonants$ [17] = "L"
consonants$ [18] = "NG"
consonants$ [19] = "R"
consonants$ [20] = "F"
consonants$ [21] = "P"
consonants$ [22] = "B"
consonants$ [23] = "F"
consonants$ [24] = "DH"
consonants$ [25] = "V"
consonants$ [26] = "HH"
consonants$ [27] = "CH"
consonants$ [28] = "G"
consonants$ [29] = "ZH"
consonants$ [30] = "JH"
consonants$ [31] = "m"


vowels$ [1] = "A"
vowels$ [2] = "I"
vowels$ [3] = "O"
vowels$ [4] = "AH1"
vowels$ [5] = "IH0"
vowels$ [6] = "AE1"
vowels$ [7] = "IY1"
vowels$ [8] = "UW1"
vowels$ [9] = "EY1"
vowels$ [10] = "OW1"
vowels$ [11] = "AH0"
vowels$ [12] = "AA1"
vowels$ [13] = "IH"
vowels$ [14] = "AE"
vowels$ [15] = "AY1"
vowels$ [16] = "U"
vowels$ [17] = "AH"
vowels$ [18] = "AY"
vowels$ [19] = "EH1"
vowels$ [20] = "OW0"
vowels$ [21] = "IH1"
vowels$ [22] = "AH2"
vowels$ [23] = "E"
vowels$ [24] = "IY0"
vowels$ [25] = "AO1"
vowels$ [26] = "OW2"
vowels$ [27] = "ER0"
vowels$ [28] = "AA2"
vowels$ [29] = "AY0"
vowels$ [30] = "IH2"
vowels$ [31] = "UH1"
vowels$ [32] = "EH0"
vowels$ [33] = "UW2"
vowels$ [34] = "UW0"
vowels$ [35] = "EH2"
vowels$ [36] = "ER1"
vowels$ [37] = "AW1"
vowels$ [38] = "IY2"
vowels$ [39] = "AY2"
vowels$ [40] = "AO0"
vowels$ [41] = "AE2"
vowels$ [42] = "AE0"
vowels$ [43] = "OY1"
vowels$ [44] = "EY2"

start = Get start of selection
end = Get end of selection
current = start
endeditor

tierOneName$ = Get tier name... 1
if tierOneName$ <> "CV"
 Insert interval tier... 1 CV
endif


textgrid = selected ("TextGrid")
select textgrid
firstInterval = Get interval at time... 2 'start'
lastInterval = Get interval at time... 2 'end'

for currentIntTwo from firstInterval to lastInterval
 currentTime = Get start point... 2 'currentIntTwo'
 currentIntOne = Get interval at time... 1 'currentTime'
 currentStartOne = Get start point... 1 'currentIntOne'
 if currentStartOne <> currentTime
  Insert boundary... 1 'currentTime'
 endif
endfor

for currentIntTwo from firstInterval to lastInterval
 currentTime = Get start point... 2 'currentIntTwo'
 currentIntOne = Get interval at time... 1 'currentTime'
 currentStartOne = Get start point... 1 'currentIntOne'
 currentLabelOne$ = Get label of interval... 1 'currentIntOne'
 currentLabelTwo$ = Get label of interval... 2 'currentIntTwo'
 if currentLabelOne$ = ""
  unknown = 1
  if currentLabelTwo$ = "sp"
   currentIntOnePlus = currentIntOne + 1
   Set interval text... 1 'currentIntOne' sp
   unknown = 0
  endif
  for k from 1 to 31
   if currentLabelTwo$ = consonants$ [k]
    Set interval text... 1 'currentIntOne' C
    unknown = 0
   endif
  endfor
  for k from 1 to 44
   if currentLabelTwo$ = vowels$ [k]
    Set interval text... 1 'currentIntOne' V
    unknown = 0
   endif
  endfor
  if unknown = 1
   printline 'currentLabelTwo$'
   printline 'currentIntTwo'
  endif
 endif
endfor

for currentIntTwo from firstInterval to lastInterval
 currentTime = Get start point... 2 'currentIntTwo'
 currentIntOne = Get interval at time... 1 'currentTime'
 currentStartOne = Get start point... 1 'currentIntOne'
 currentLabelOne$ = Get label of interval... 1 'currentIntOne'
 currentLabelTwo$ = Get label of interval... 2 'currentIntTwo'

 nextIntOne = currentIntOne + 1
 nextStartOne = Get start point... 1 'nextIntOne'
 nextLabelOne$ = Get label of interval... 1 'nextIntOne'
 if currentLabelOne$ = nextLabelOne$
  Remove boundary at time... 1 'nextStartOne'
 endif
endfor

Op 25 feb 2012, om 16:11 heeft latinista90 het volgende geschreven:

> > Column "label" does not exist.
> Script line 22 not performed or completed:
> « Sort rows... label »
> Please change something in the window `Formant plotter', or click Cancel in
that window.
>
> The problem is related to the file which I have to give as path, but I don't
now exactly what it does mean, that a "column does not exist".

In your script you told Praat to sort the rows of a Table alphabetically. You
told Praat to use a column named "label" to do the sorting. For instance, if
column "label" contains "b" in row 1, "c" in row 2, and "a" in row 3, then row 3
will go to the top and row 2 to the bottom.

Apparently your Table does not have a column named "label".
_____

Paul Boersma
University of Amsterdam
Linguistics
Spuistraat 210, room 303
1012VT Amsterdam, The Netherlands
http://www.fon.hum.uva.nl/paul/