Thanks Ethan! I'm going to post this in my Fall 20 Linguistics Phonetics class. The book (Ladefoged and Johnson) has a whole section of a chapter on ToBI which is just way too compressed and poorly explained for undergraduates. So, I'll move over to having them just do Ls and Hs (no * ! ) based on these beautiful contours.
The green values represent the intensity (in dB) at the point in time where you last clicked in the spectrogram. If you haven't clicked anywhere, the value is at the middle point in time. If you highlight a section, the value changes to the mean value for the section you have highlighted. You can watch this mean value increase and decrease as you highlight a section of the spectrogram.
Yes, you definitely can. If you have created a pitch object in Praat, then, from the main menu, select the Pitch object and choose Sound, and then To Sound. You can choose to render the pitch contour as either pulses, hum, or sine wave.
Hi! great tutorial! thanks! any idea why it would only work for me with pitch and NOT intensity? (when I select intensity object + textgrid all the buttons disappear!).
Interesting, my buttons disappear too! I don't think I ever tried doing this with intensity alone, so I never noticed. I think you can still get something usable by e.g. plotting the text by itself, and then highlighting an area and drawing the intensity there, but it's not quite the same. You could try asking here: groups.io/g/Praat-Users-List
Hi. Thanks a lot for this helpful video. I just have a question. I have the words under the pitch contour in the drawing box and not above the line as u did. what is the think I miss here to do the same? thanks
Hi Amira. I'm not sure. I made this video 4 years ago, and Praat may have changed since then. I haven't tried this recently, so I can't say. I would suggest trying the Praat Users list: groups.io/g/Praat-Users-List
Sure, you can do that. There are several ways to do it. Say you want F0 measurements, you can view a sound file, select the section you want to analyze, and then in the "Pitch" menu choose "Show Pitch" and then "Pitch listing". This will open a text box with time and F0 data. If you have many files, you can also script this, so that you can calculate and export your data automatically.
Hmm... I think you are asking about TOBI coding, which is something I really don't know anything about. But, you can click anywhere in the waveform or spectrogram and get the f0 values, if that helps.
Hello Ethan, thank you for your tutorial. Could I ask you how can I decide the intensity setting range? and Also the pitch setting range? I am studying leopard calls
Wow, studying leopard calls sounds very cool! I think if you load a sound file, go to View & Edit, and then look in the Intensity and Pitch dropdown menus you will find what you are looking for under "Pitch settings..." and "Intensity settings..."
Hello Ethan Weed. Thanks so much for sharing this! I have few questions regarding speech analysis; 1. Is PRAAT the best open source software for singing sound (without instrument) analysis? 2. There are few papers suggest to perform MFCC to obtain envelope, does MFCC really represent amplitude envelope? If yes how can I extract envelope from PRAAT? 3. Should I consider music audio features (eg. melody, tempo etc) for singing sound analysis? Or just focus on speech features? Thanks in advance!
Hi, I can't tell you what the best open software for voice analysis is, but Praat is certainly a good option. It is reasonably user-friendly, and is well-known. Mel-frequency cepstral coefficients (MFCCs) are a measure of spectral qualities of the signal (Fourier-transform the signal, convert to mel scale, take the log, and cosine transform (see e.g. en.wikipedia.org/wiki/Mel-frequency_cepstrum and en.wikipedia.org/wiki/Cepstrum). I do not think this is what you need if you are looking for the amplitude envelope. I have not tried this personally, but I can see that He and Dellwo have published a Praat script that finds the amplitude envelope here: www.zora.uzh.ch/id/eprint/127258/1/HeDellwo_Interspeech2016.PDF You should be able to copy/paste their script into the script editor in Praat, and (perhaps with a few changes) use it to work with your files. I cannot answer your last question, as this depends entirely on what your goal with the analysis is. Hope this helps, Ethan
A really helpfull video. Thanks a lot! I followed the same process as you but I cannot take the same picture of pitch as you (a line of points). I only get a messed up mixure of lines. Any suggestions?
Hmm. Maybe Praat is having a hard time finding the F0 in your sound file. You could try with a file from talkbank.org and see if you can get that to work first.
@@ethanweed After some restarts it seems that it's working. I don't have the same depiction as you but now it seems reasonable. What version of Praat do you use?
@@gabrielkaravasilis2209 Good to hear it is working now. I just update to whatever the most recent version of Praat is. I made that video several years ago now, so I'm not sure what version I was using at the time.
I'm really not the right person to ask, I haven't done many of my own recordings.But it really depends on your needs, which I know is a non-answer! Depending on what you want to do with the recordings, I could imagine anything from a fancy microphone in an anechoic chamber to a smartphone on a table being appropriate.
Hi there, when I view pitch analysis I don't get a smooth line like yours. Instead it's all choppy. Do you have any idea what I should adjust the settings to or what settings to adjust? B/c I'm not remotely technical I can't find it myself thus far.
You have just helped me a lot! Thank you!
Ethan Weed Thank you for this video! I'll be using this app for my research study.
What a brilliant tutorial, thank you! ;)
Wow, this is a phenomenal tutorial, thanks!
Thanks Ethan! I'm going to post this in my Fall 20 Linguistics Phonetics class. The book (Ladefoged and Johnson) has a whole section of a chapter on ToBI which is just way too compressed and poorly explained for undergraduates. So, I'll move over to having them just do Ls and Hs (no * ! ) based on these beautiful contours.
Thanks for this comment, it made my day!
5:08 you are my hero thank you
Thank you very much, it was quite informative for me.
thank you so much!!! this is exactly what i need for a project i'm doing! :D
Great, I'm glad it was useful :-)
how did u select the sound and textgrid at the same time please? Mine just doesn't work when I click on the sound.
On a Mac I do command-click. I think it is ctrl-click on Windows.
Hi! I have a question: the green value of the intensity of the sound that appears on the screen rapresents the medium value of it?
The green values represent the intensity (in dB) at the point in time where you last clicked in the spectrogram. If you haven't clicked anywhere, the value is at the middle point in time. If you highlight a section, the value changes to the mean value for the section you have highlighted. You can watch this mean value increase and decrease as you highlight a section of the spectrogram.
Thanks for your video! Is it possible to save pitch file as a sound?
Yes, you definitely can. If you have created a pitch object in Praat, then, from the main menu, select the Pitch object and choose Sound, and then To Sound. You can choose to render the pitch contour as either pulses, hum, or sine wave.
@@ethanweed thank you
Hi! great tutorial! thanks! any idea why it would only work for me with pitch and NOT intensity? (when I select intensity object + textgrid all the buttons disappear!).
Interesting, my buttons disappear too! I don't think I ever tried doing this with intensity alone, so I never noticed. I think you can still get something usable by e.g. plotting the text by itself, and then highlighting an area and drawing the intensity there, but it's not quite the same. You could try asking here: groups.io/g/Praat-Users-List
Hi. Thanks a lot for this helpful video. I just have a question. I have the words under the pitch contour in the drawing box and not above the line as u did. what is the think I miss here to do the same? thanks
Hi Amira. I'm not sure. I made this video 4 years ago, and Praat may have changed since then. I haven't tried this recently, so I can't say. I would suggest trying the Praat Users list: groups.io/g/Praat-Users-List
Thank you so much!!
Is it possible to export data from Praat? this is needed to calculate the standard deviation in IBM SPSS Statistics.
Sure, you can do that. There are several ways to do it. Say you want F0 measurements, you can view a sound file, select the section you want to analyze, and then in the "Pitch" menu choose "Show Pitch" and then "Pitch listing". This will open a text box with time and F0 data. If you have many files, you can also script this, so that you can calculate and export your data automatically.
Thanks, How do we measure contours, H-L* or L-L?
Hmm... I think you are asking about TOBI coding, which is something I really don't know anything about. But, you can click anywhere in the waveform or spectrogram and get the f0 values, if that helps.
Hello Ethan, thank you for your tutorial. Could I ask you how can I decide the intensity setting range? and Also the pitch setting range? I am studying leopard calls
Wow, studying leopard calls sounds very cool! I think if you load a sound file, go to View & Edit, and then look in the Intensity and Pitch dropdown menus you will find what you are looking for under "Pitch settings..." and "Intensity settings..."
Hello Ethan Weed. Thanks so much for sharing this! I have few questions regarding speech analysis;
1. Is PRAAT the best open source software for singing sound (without instrument) analysis?
2. There are few papers suggest to perform MFCC to obtain envelope, does MFCC really represent amplitude envelope? If yes how can I extract envelope from PRAAT?
3. Should I consider music audio features (eg. melody, tempo etc) for singing sound analysis? Or just focus on speech features?
Thanks in advance!
Hi,
I can't tell you what the best open software for voice analysis is, but Praat is certainly a good option. It is reasonably user-friendly, and is well-known.
Mel-frequency cepstral coefficients (MFCCs) are a measure of spectral qualities of the signal (Fourier-transform the signal, convert to mel scale, take the log, and cosine transform (see e.g. en.wikipedia.org/wiki/Mel-frequency_cepstrum and en.wikipedia.org/wiki/Cepstrum). I do not think this is what you need if you are looking for the amplitude envelope.
I have not tried this personally, but I can see that He and Dellwo have published a Praat script that finds the amplitude envelope here: www.zora.uzh.ch/id/eprint/127258/1/HeDellwo_Interspeech2016.PDF You should be able to copy/paste their script into the script editor in Praat, and (perhaps with a few changes) use it to work with your files.
I cannot answer your last question, as this depends entirely on what your goal with the analysis is.
Hope this helps,
Ethan
how do I select two simultaneously？
On a Mac I do command-click. I think it is ctrl-click on Windows.
thank you！it worked！@@ethanweed
A really helpfull video. Thanks a lot!
I followed the same process as you but I cannot take the same picture of pitch as you (a line of points).
I only get a messed up mixure of lines. Any suggestions?
Hmm. Maybe Praat is having a hard time finding the F0 in your sound file. You could try with a file from talkbank.org and see if you can get that to work first.
@@ethanweed After some restarts it seems that it's working. I don't have the same depiction as you but now it seems reasonable. What version of Praat do you use?
@@gabrielkaravasilis2209 Good to hear it is working now. I just update to whatever the most recent version of Praat is. I made that video several years ago now, so I'm not sure what version I was using at the time.
@@ethanweed I see. Thanks again!
Thank you so much for sharing this
I need it for my thesis
may I ask how do I add annotation and intensity without drawing the pitch contour?
You can just skip the step at 2:00 where I click on "show pitch", and you should be all set.
You rock ! thanks
:-)
which recorder do you use to record your data?
I'm really not the right person to ask, I haven't done many of my own recordings.But it really depends on your needs, which I know is a non-answer! Depending on what you want to do with the recordings, I could imagine anything from a fancy microphone in an anechoic chamber to a smartphone on a table being appropriate.
Hi there, when I view pitch analysis I don't get a smooth line like yours. Instead it's all choppy. Do you have any idea what I should adjust the settings to or what settings to adjust? B/c I'm not remotely technical I can't find it myself thus far.
I've just seen the comment. You have to set the range to 75 to 300 for men and 100 to 500 for women
Thanks you
thank you
You're welcome, glad you found it useful :-)