Table of contents
Questions for Chapter 1
1.1 Acoustic phonetics Top
The study of acoustic phonetics was greatly enhanced in the late 19th century. Acoustic phonetics is a subfield of phonetics which investigates the acoustic aspects of speech sounds, like the amplitude of a waveform, its duration, its fundamental frequency, and other properties of its frequency spectrum. Learning acoustic phonetics is of great significance for language teachers in their pronunciation teaching and speech therapists in assessment and planning of remediation.
Key concepts in acoustic phonetics (Adapted from Yavas, 2011:101)
There are four acoustic properties of speech sounds: frequency, time, amplitude, and formant.
a. Frequency: Frequency relates to the individual pulsations produced by vocal cord vibrations for a unit of time. The rate of vibration depends on the length, thickness, and tension of the vocal cords, and thus is different for child, adult male and female speech. A speech sound contains two types of frequencies: fundamental frequency (F0) which relates to vocal cord function and reflects the rate of vocal cord vibration during phonation (pitch) and formant frequency which relates to vocal tract configuration.
b. Time: Time as a property of speech sounds reflects the duration of a given sound.
c. Amplitude: The amplitude is marked by darkness of the bands: the greater the intensity of the sound energy presents in a given time and frequency, the darker will be the mark at the corresponding point on the screen.
d. Formant: A formant is a concentration of acoustic energy around a particular frequency in the speech wave. There are several formants, each at a different frequency, roughly one in each 1000Hz band. To put it differently, formants occur at roughly 1000Hz intervals. Each formant corresponds to a resonance in the vocal tract.
By convention, oral formants are numbered consecutively upwards from the lowest frequency. All vowels can be characterized by F1 and F2. For example, in the following spectrogram Figure 1.1 for the vowels in ‘hit’ (/ɪ/) and ‘hot’(/ɒ/), we can find that F2 is higher (close to F3) in the first vowel (/ɪ/), but lower (close to F1) in the second vowel (/ɒ/). Vowels traditionally known as "front vowels" have F1 and F2 a good distance apart, like the first vowel /ɪ/ here. Vowels traditionally known as "back vowels" have F1 and F2 so close that they touch each other, like the second vowel/ɒ/ here.
1.2 Introduction of Praat Top
Praat is a freeware program for the analysis and reconstruction of acoustic speech signals. You can analyze, synthesize, manipulate speech and create high-quality pictures for your articles and thesis with it. It can be downloaded from the following website: http://www.fon.hum.uva.nl/Praat/.
1.2.1 What is Praat? Top
Praat is an open-software tool for the analysis of speech in phonetics. It was designed, and continues to be developed, by Paul Boersma and David Weenink of the University of Amsterdam. It's free and available for most platforms.
Praat was designed to cater for different needs with easy interface, many default options to learn by trying, searchable manual, and various possibilities of analysis, manipulation and labeling. (Goldman, 2004: 1)
1.2.2 What we can do with Praat? Top
With Praat, you can …
l generate waveforms, wide and narrow band spectrograms, intensity contour and pitch tracks;
l make recordings, edit a recorded sound, and extract individual sounds for further analysis;
l get information about pitch, intensity, formants, pulses and etc;
l enhance certain frequency regions; segment and label words, syllables, or individual phonemes;
l put your work in graphic form ('draw a plot') for printing.
Once you've opened Praat, several windows will open automatically, and there are many other windows which will pop up later on when you click different buttons, so we’d better discuss different windows in Praat before we introducing different buttons.
a. Praat Objects window
The Praat Objects window (Figure 1.3 on the left) is where you can open, create and save files. This menu can be used to open the various editors and queries which you’ll need to work with sound files.
Figure 1.3 Praat Objects Window
After opening the program, Praat has no objects in its object list. Therefore, the list is empty and the buttons (at the bottom) are disabled and shown in grey. When you put the sound files in the list, menus and buttons are become dynamic, and they may change (appear, disappear or be disabled) according to the selected objects.
You can create
a new sound via Menu: 'New'
Record a mono Sound
to the list
the file nameà
Then, the sound you’ve just recorded will appear is now in the list of objects with the file name as “Sound+ name”. As it is selected (in blue), many buttons appear on the right and at bottom are now enabled.
The five buttons at the bottom are called the fixed buttons. They might be disabled if nothing is selected or present in the list. These functions are common to all objects whatever the type.
Ø Rename: Rename an object
Ø Copy: Duplicate an object
Ø Remove: Delete an object
Ø Info: Get the results from various queries on an object
Ø Inspect: Browse the internal data of an object
The buttons on the right of the list (in Figure 1.4) are the dynamic buttons. We'll go through most of them in our following session 1.3. The buttons on the right won't appear if no object is selected or several objects of different types are selected at the same time.
Figure 1.4 Buttons on Praat Objects Window
b. Editor window (Adaped from Will Styler, 2014:7).
The Editor window (Figure1.5) is where you’ll spend most of your time processing and measuring the sound file. You can access the Editor window by selecting a sound and clicking on"View & Edit". When you open the Editor window, the sound’s waveform and spectrogram will be shown on the top and the bottom respectively, and the cursor will allow you to make selections and measurements. The menus on the top allow you to show and hide different acoustic information (e.g. formants, pitch, intensity), as well as to make more detailed queries.
Figure 1.5 Editor Window
The menus on the top of the Editor window contain the following options( Lieshout, 2005: 13-14):
• File: to extract selections in different ways, to open a script file, etc.
• Edit: to copy or paste parts of a signal, etc.
• Query: to get information on the cursor position, selection boundaries, define settings for logs and reports, etc.
• View: to select the contents of the window (spectrogram, intensity etc.) and control zoom settings.
• Select: to control cursor positions.
• Spectrum: to control the spectrogram settings and extract information; the frequency value at the cursor position is indicated on the left hand outside of the panel in a red font.
Figure 1.6 Spectrogram menu
• Pitch: to control the pitch settings and extract information; by default the pitch signal is shown in a bright blue solid line and the value at the cursor position is indicated on the right hand outside of the panel in a dark blue font.
Figure 1.7 Pitch menu
Figure 1.8 Pitch contour overlained on spectrogram in Praat
• Intensity: to control the intensity signal settings and extract information; by default the intensity signal is shown in a yellow solid line and the value at the cursor position is indicated on the right-hand side of the panel in a bright green font.
Figure 1.9 Intensity menu
Figure 1. 10 Intensity contour in Praat
• Formant: to control the formant settings and extracts information; by default the formants are shown in red dotted lines. You can adjust the numbers of formants presented and the maximum formant via ‘Formant’à ‘Formant settings…’In later chapters (1.4.5), we will introduce how to adjust the formant settings and extract formant information in detail.
Figure 1.11 Formant menu
Figure 1.12 Formants in Praat
• Pulses: to set pulses and to extract specific information on voice parameters like jitter and shimmer; pulses are indicated in the top panel with vertical blue solid lines.
Figure 1. 13 Pulses in Praat
c. Info window
When you make a query, either in the editor window or from the objects window, the Info window will pop up with your results. You can also print to this window when scripting in Praat. Please note that information printed here will not necessarily be saved, and running a new query will overwrite it by default.
Figure 1. 14 Praat Info window to show information on mean pitch
d. Praat Picture window
The Praat Picture window (Figure 1.15) is used to create and display publication-quality images, and is open by default when you start the program.
Figure 1. 15 Praat Picture window
1.3 Operations in Praat Top
1.3.1 Starting Top
You can go to website http://www.fon.hum.uva.nl/Praat/download_win.html and download the Praat to your computer . To start up the Praat program, just double-click this icon .
When you run the program, you will see the two windows below. As we mentioned before, the left window is the “Praat Objects” window. On the left-hand side you will see the list of your speech files. These can either be created from scratch or read from a file. Additionally, the right window is the “Praat Picture” window which is used to plot graphs. When you select a sound for further analysis and click button ‘View&Edit’, the Editor Window will appear, where you’ll spend most of your time,
1.3.2 Recording Top
When you want to make recordings in Praat, the first thing you need to do is to connect a high-quality microphone to the MIC input of your computer, and then choose “New” on the top menu and select “Record mono record” , and the Sound Recorder window will appear.
Figure 1.16 The Sound Recorder window
Set the sampling rate to 22050Hz which is the most sufficient, and then take a deep breath and click the “Record” button to start your recording.
When you speak, the meter will show the input level with green bars. When you finish, click the “Stop” button. You can use the “play” button to hear what you have recorded and repeat steps 2 and 3 until you are satisfied with your recording.
If the recording is to your satisfaction, you can add a name for the recording in the blank after “Name” and click on the “Save to list & Close” button. This will put your recording in the “object window”.
1.3.3 Opening Existing files Top
Apart from recording a new sound from a microphone, you could read an existing sound file from your disk. Click “Read from file” from the menu “Open”, Praat will read the files from your computer.
1.3.4 Editing Top
Praat offers a lot functions to visualize, play and extract information from a sound object. You can start by selecting the speech object and then choosing ‘View&Edit’ from the main menu on the right-hand side of the “Objects window”. The new "Editor window", which has been introduced in detail in 1.2.3 b, will appear.
On the pull-down menu of Edit, you can find the following functions
Ø Cut: cut the selection to the clipboard.
Ø Copy selection: copy the selection to the clipboard.
Ø Paste: paste the clipboard contents to the cursor.
Ø Set selection to zero: set the selected samples to zero.
Ø Reverse selection: reverse the selected part of the sound.
Remarks: Only when you select a part of the recording, all the functions in the pull-down menu can be activated, or they are grey and unavailable.
You can move the red dash line to change the scope of recording. If you want to cut, copy, and paste between the sounds, you can open more than one sound, and then select “cut”, “copy”, and “paste” between the sounds by moving a selected part of the sound to another location, and using “cut” and “paste” from the “Edit” menu.
You can find the total duration of the recordings at the bottom and the duration of your selected part below the spectrum. To play the selected part of the sound, you only need to click on the rectangle below it.
1.3.5 Drawing Top
The easiest way to save a Praat picture is to take a screenshot of it. However, if you want to create and annotate publication quality graphs, more efforts are required. The standard way to print that spectrogram on paper, or to save it as an image file for inclusion in a report or presentation, is to transfer it to the Picture window with the "Draw" function from the Objects window.
You can draw a picture in Praat by the following steps:
1. First, find the Picture window. If you accidentally closed it, you can just reopen Praat.
2. Determine the physical size of the plot by changing the selection in the Praat "Picture window" (pink rectangular shape) before you draw the graph. Click in an area of the Praat “Picture window” (e.g., left upper corner) and draw the new shape while keeping the left-mouse button pressed。
3. Choose the given recording file on the left-hand side of Praat objects and click the “Draw” button. Set these parameters as you’d like. Figure 1.26 shows the spectrogram painted to the Praat Picture window
Figure 1. 24
Figure 1. 25
1.3.6 Garnishing your pictures Top
Now, we will introduce some basic skills about how to garnish your pictures using Margins, World, Pen, and Font in upper menu to further garnish your picture.
Margins: You can use “Margins” to add the texts and marks in your picture. Let’s use an example to illustrate the functions.
When you click "Marks left every", the following window appears.
In this window,
· Units is the basic unit for this particular scale; in this example it is 1.0
· Distance is the interval at which something is to happen (numbers, ticks, lines);
· Units and Distance work together, such that Distance is a fraction or multiple of Units; Distance is set to 0.1 for this example; note that Praat insists you enter 0 before numbers smaller than 0.
· Numbers writes the appropriate number at locations defined by Units and Distance; Numbers is not selected for now
· Draw ticks places tick marks at the defined locations
· Draw dotted lines draws a dashed line through the diagram at each defined location
Following these additions, the diagram looks like this:
Figure 1. 29
Similarly, you can add ticks and numbers at the bottom by opening the Margins menu and selecting Marks at bottom every, the dialog opens:
Then click OK, the diagram looks like Figure 1.31
If you want to add titles at the top, you can use Text top command in the Margins menu.
When you click Text top, the following window will pop up, and then put tick before "Far" and type in the text you want to put at the top, for example “good morning”, and then click "OK"
Remarks for 'Far': Two lines can be written at the top, one at the edge of the selected area and one just inside. You can decide which line you want to use by ticking the "Far" selection. When Far is ticked, you are going to place the text at the very edge of the drawing area. This is the first line. If you want to place the next line just inside, you can leave 'Far' unchecked and type the text you want, for example, we type "recording 1" in Figure 1.35.
After filling in the text for both lines, you can see two lines on the top of the pictures.
Pen: You can also use Pen to decide the line types and colors of your diagram. For example, I set the line as Dotted line and Blue . Then, the diagram will become like Figure 1. 38
(This part is adapted from Wood’s beginners' guide to Praat (2005), which is available at http://person2.sol.lu.se/SidneyWood/Praate/printediting.html#Editing lines)
Besides drawing waveforms, Praat can also draw the spectrum, pitch contour, formant, intensity contour, etc.
Here, this manual will demonstrate how to draw a spectrum step by step in Praat.
1. Open the Editor window.
2. Determine the physical size of the plot by changing the selection in the ‘Praat picture’ window (pink rectangular shape) before you draw the graph.
3. Click in an area of the Praat "Picture window" (e.g., left upper corner) and and draw the new shape while keeping the left-mouse button depressed.
3. Click the Spectrum and find "Paint visible spectrogram", then click it.
4. When the Paint window pops up, just click OK, and the spectrum will be drawn.
Figure 1. 41
Similarly, you can draw the formant contours by using the same steps. If you still select the same area in Picture window, the formants that are newly drawn will overlap with the old spectrum (Please see Figure 1. 43)
Figure 1. 42
The color of the formant presentation can be adjusted in Pen
Praat can save what you currently select, so make sure you have selected what you want. Before we save our diagram we should know how many formats we can use . When you click "File", you can see the following three formats:
Ø Praat picture file: This is a file format that is exclusive to Praat and cannot be opened or edited by any other program; the entire contents of the drawing board are saved as one image; the images are saved as a ".prapic" file. You will notice that this is the only image format that Praat can open again (via "Read from Praat picture file"), so you can use this format when you wish to continue working on your page in the Picture window on a future occasion.
Ø Windows meta file: The enhanced metafile is a standard Windows file format: use it if you wish to open your work in some other Windows program such as an image editor like Photoshop or word processor; only the selected area is seen in the saved image, so make sure you select what you want before saving; the images are saved as ".emf" file. When you double click the file, the file will be opened in your default image presentation software.
Ø EPS file: This is a postscript file, for computers that are set up to use Postscript for image editing and for printing; only the selected area is seen in the saved image, so make sure you select what you want before saving; the images are saved as .eps
(This part is adapted from Sidney Wood (2004)
"Praat for beginners" that is available at
In this manual, we only demonstrate how to save the picture files as .eps, you can later convert it to "Windows meta file" or other formats that suit your needs.
You can insert
an ‘eps’ file into your document from the insert pull-down menu at the top of
the Microsoft Word window. Then click on the "Insert Picture" dialog box. Choose
"Encapsulated PostScript (*.eps)" for "Files of type." Navigate to the
directory that contains the EPS file, and double-click on the filename of
the EPS file that you want to import.
Now let’s save the picture file as an EPS file.
First, select the picture you want to save and click Save as EPS file, and save the file on desktop as “Praat_picture 1”
Second, open your word file, click insertàPictureà Choose "Encapsulated PostScript (*.eps)" for "Files of type."à Navigate to the directory that contains the EPS file, and double-click on the filename of the EPS file that you want to import.
Then the picture made in Praat will be inserted in your word file as shown below.
1.4 Analysis of speech in Praat Top
Before using Praat to do sound analysis, we have to be clear about know that what information we can get from Praat. Table 1 presents some major acoustic variables we usually use to analyze the speech sounds. (See Figure 1. 49) for visual presentation of the variables.
1.4.1 Extracting sound Top
If you want to extract a section of a sound (usually a single word or vowel) into a different Sound object for analysis, you can
1. Select the section of sound by cursor
2. Click File à Extract Selected Sound (preserve times)
By doing this, a new sound file will be is created in the Objects window, containing just the selected part of the original sound.
1.4.2 Spectrogram Top
a. Showing the Spectrogram
Normally the waveform and spectrogram will be presented automatically if you select one file and click ‘View and Edit’ as Figure 1.50
b. Adjusting the Spectrogram settings
The most important settings here are the window length and view range.
i. View range (Hz).
View range decides how much of the spectrum is shown. For speech, we normally set the range from 0 to 5000 or 6000 Hz, but for examining fricatives, we might need to set it as high as 15,000 Hz. For music, we may need to focus on the area from 100 to 2000 Hz. (Revised from Styler, 2012)
You can adjust the View range by clicking Spectrum àSpectrogram Settings
ii. Window length
Praat can provide you with both Broadband spectrogram and Narrowband spectrogram by adjusting the window length. The shorter the window length, the larger its bandwidth (Bandwidth=1.299/ window length). There is no clear cut boundary between Broadband spectrograms and Narrowband spectrograms, if the window length is around 3-5 ms(bandwidth:200-300Hz), the resulting spectrogram is called "wideband". For the window length around 20-30m s (bandwidth:30-50Hz), the spectrogram is called "narrowband". Wideband spectrogram is used to observe the formant structure while narrowband spectrograms reveal the harmonic structure (pitch information).
l Broadband spectrogram (Window Length： 0.005s) is used to observe the formant structure of sound，and it is the default setting in Praat. (See Figure 52 )
l Narrowband spectrogram (Window Length：0.025s ) can be used to look at the harmonics structure (F0/Pitch information) (Figure 53)
You can adjust the window length by clicking ‘Spectrum’ à ‘Spectrogram Settings’ à set the ‘Window Length’ to 0.025s (or the narrowband window length of your choosing) à Click OK
Now, you can see harmonics clearly in this narrowband spectrogram.
If you set the view range roughly as 0 - 500 Hz for speech in this narrowband spectrogram, the contours of the harmonics will accurately represent the pitch contours of the voice, which can give you a sense of the pitch (F0) contour before using the Praat pitch tracker for more precise measurement.
To return to a broadband spectrogram, you can click ‘Spectrum’ à ‘Spectrogram Settings’ àSet the Window Length to 0.005 (or the broadband window length of your choosing) à Click OK
And then you’ll be back to the default broadband spectrogram.
Before we illustrate how to measure pitch in Praat, let’s discuss what the pitch is and what it used for.
Pitch is a term used to refer to variations in fundamental frequency (F0), which serves as an important acoustic cue for tone, lexical stress, and intonation. For example, in Chinese, which is a tone language, each syllable or morpheme may have its own pitch
a. Extracting information about pitch
1. Display the pitch track: PitchàShow pitch
2. At this point, a blue line will be placed on the spectrogram representing the pitch. At this time, you can place the cursor at the point and read the blue number on the right side of the window.
3. Or you can position the cursor in a stable middle part of the blue track and click “Pitch “and then select “Get pitch”. A local pitch value will be displayed in a separate window.
b. Getting Maximum, Minimum, and Average pitch for a section of speech
1. Select the portion of the sound for which you would like the Maximum, Minimum or Average Pitch
2. Select the proper command for your task from the top menu: PitchàGet Pitch/Get Maximum Pitch/Get Minimum Pitch
c. Improving the pitch contour by adjusting the pitch settings
Sometimes you will find that the blue pitch contour jumps up and down, doubling and halving the actual F0, and in many cases, especially where the speaker is creaky, the pitch track will drop out altogether, which is because Praat’s default pitch range is not appropriate for the file you’re analyzing. Therefore, in order to make the pitch track more visible and better reflect the speaker's voice , you may need to adjust some of the pitch settings via Pitch à Pitch settings (see Figure 1.58)
The fundamental frequency of the voice (pitch) usually ranges from approximately 30–300 Hz, but this varies according to different speakers: typically males’ pitch ranges from 50-180Hz and females from 80-250Hz, so we usually set the pitch range to a reasonable range of 50 - 400 for general usage.
If you have a general sense on what the speaker's actual range is (e.g. getting from the previous measuring), you can set the minimum to just under the speaker's lowest F0 and the maximum to just over their highest pitch excursion.
If the pitch contour is too low in the spectrogram, you can increase the maximum value of the pitch range (e.g. increase from 400 to 500); if the pitch contour is too high, you can decrease the maximum value of the pitch range (e.g. increase from 400 to 300).
(This part is adapted from Stonham's lecture notes (p.13) that is available at http://stonham.dyndns.org/phonetics/handouts/prosod_hndt.pdf)
On the right side of the window, you can find the fundamental frequency (F0), which is marked in blue, while on the left side, the frequency value marked in red is formant frequency
Figure 1. 59
1.4.4 Intensity Top
1. Position the cursor in a stable middle part of the sound and do the following
2. Go to “Intensity” and select “Get intensity”. A local intensity value will be displayed in a separate window.
Figure 1. 60
1.4.5. Formant Top
Let’s discuss how to extract information about formant values
1. Position the cursor in a
stable middle part of the sound.
1. It’s more efficient to use ‘Editor’ ‘Formants’ ’Formant Listing’, which will give you values for F1, F2, F3 and F4, along with the time point at which the measures were taken.
2. Adjust the Formant settings to make the measure more accurate.
You can go to ‘Formant’ and select Formant settings
n For the male, set the maximum formant(HZ) as 5000Hz
n For the female, set the maximum formant(HZ) as 5500Hz
n For the children, set the maximum formant(HZ) as 8000Hz
Questions for Chapter 1
1. Try to record a single word or phrase, add a name and save it to your computer.
2. How do you visualize the sound?
3. What does the blue contour stand for in the spectrogram?
4. How do you display the pitch, formant, intensity respectively in Praat?
5. How do you open an existing file in Praat?
6. How do you get the information of the duration of selected part of sounds or the whole recording?
7. How could you get the information about the “pitch”?
8. How do you cut, copy, and paste between the sounds?
9. How do you adjust the pitch contour if the pitch contour is too high or too low?
10. How do you draw a picture in the “Picture” window?
(See the suggested answers at the end of the manual)