![]() |
Examples
|
em
português
Examples
To
decrease the total download size of Sound Ruler, the sounds used in these examples
have been saved with low sampling rate and 8 bits. The default settings should
work fine with these sounds.
When analyzing your own sounds, you will probably work
with 16 bits and higher sampling rates. It is worthwhile taking the time to
adjust your settings and find the best compromise between precision and speed.
Then you can save them (menu "Options"-"Settings"-"Save")
to make them available for the next analysis.
MEASURING SOUND
GRAPHING
TOOLS
MEASURING SOUND
Easy
calls
1) Go to the menu "File"-"Open",
and open the example file P_streckeri.wav from the <soundruler>\sounds
directory.
2) Enter 4 in the box that asks about the length
that each section should have.
3)
The file opens and its information is posted on Sound Ruler's title bar.
4) The oscillogram of the first Section is plotted
in the top axis.
5) Click on the Auto button by the Section list.
The calls are automatically recognized.
6) Sound Ruler focuses on the first call and plots
an oscillogram, spectrogram and power spectrum of it. It also takes all the
measurements and displays them in the results list. Don't panic when you see
the results list! The names of the quantities are really compressed, but if
you click on one, a more extensive description shows up on the status bar, at
the bottom of the screen. If you want to examine descriptions for several variables
at once, click on the describe button above the results list. The descriptions
will be pasted into the TextPad.
7) Click on the Next button by the Calls list to move the focus to the second call. Or click on the Calls list to move strait to the call that you want.
8) To save your results, go to "File"-"Save data" and all calls analyzed will be available in a spreadsheet.
9) To get an idea of the precision of the measurements, repeat the analysis with the file cca2.wav from the <soundruler>\sounds directory. It has synthetic sounds that start at 1, 2, and 3 seconds, have a duration of 0.4 seconds and frequencies of 1, 2, and 4 kHz.
Filtering
1) Go to the menu "File" -"Open",
and open the example file P_ornata.wav from the <soundruler>\sounds directory
(see step 1 of the example " easy calls").
2) Enter 4 in the box that asks about the length that
each section should have (see step 2 of the example " easy
calls").
3) The file opens and the oscillogram of the first
Section is plotted in the top axis. (see step 3 and 4 of the example "
easy calls").
4) Click on the Erase Filter button, over
the Spectrogram. The section plot is updated.
5) Play the sound and notice how noisy this recording
is. This is because of wind.
6)
Do some Manual or use Sample sound recognition to get the Spectrogram plotted.
For Manual sound recognition, left-click (green bars will apprear) at the beginning
and end of one call. Then click Manual.
For Sample sound recognition,
left-click (green bars will appear) at the beginning and end of one call and
once more before the next call. Then click Sample.
7) Observe in the call oscilogram how noise affects the sound recognition.
8) Now click on the Spectrogram, above the base line, but bellow the frequency of the call.
9) Click on the Update Filter button. The Section
is plotted again, with the filtered signal.
10) Now do some call recognition (repeat step
6) . Much cleaner, huh?!?
If you liked it, now click Next by the Section
list and repeat the procedure, to see clean sounds emerge from the wind...
NOTE: Right-click to determine a maximum above
which energy is filtered out.
NOTE: Remember to erase the filter every time
you start with a new kind of signal, to make sure that you do not loose part
of the signal.
NOTE: The filter exhibits some gradation on its
behavior at the critical frequency (red line). If you click too close to the
signal you are interested in, it will loose some energy.
Trilled
calls
1) Go to the menu " File" -"Open"
, and open the example file H_versicolor.wav from the <soundruler>\sounds
directory. (see step 1 of the example " easy calls").
2) Enter 4 in the box that asks about the length
that each section should have (see step 2 of the example " easy
calls").
3) Now we want to help Sound Ruler not to get confused
with the pulsing. Go to the menu "Options"-"Section"
and set the "minimum silence to end" to a value that is longer that
the inter pulse interval (for this example 150 will work).
4) Click on each side of the call, once on the
next call and Sample (see step 6 - Sample of the example " filtering").
5) For some files, you might have to repeat this
steps a few times to find the best balance between recognizing what you don't
want and rejecting what you want. You can also play with the critical amplitude
and the among of sound included around the call.
6) If you loose your patience, click on each side
of each call and Click on the Manual button. It does not matter because measurements
are only taken in the next step (see step 6 - Manual of the example " filtering").
7) Now lets repeat for the call what we did in
the section. Delimit a pulse, click on the next one and then on Sample. Each
pulse is now recognized and measured.
A
hard one
1) Go to the menu " File"-"Open"
, and open the example file P_clarkii.wav from the <soundruler>\sounds
directory (see step 1 of the example " easy calls").
2) Enter 4 in the box that asks about the length that
each section should have (see step 2 of the example " easy
calls").
3)
Do a Sample call recognition (see step 6 - Sample of the example " filtering").
4) Notice that the weakest pulses are not recognized in the call oscilogram graph
.
5) Here
are your options to solve this problem:
a- Use intelligent
recognition to account for the gradual change in pulse amplitude. Go to the
menu "Options"-"Call" and change the "Expectation for
next pulse" to a higher level of complexity.
b- Reduce the minimum
amplitude for recognition.
c- Click on each
side of the pulse that you want to incorporate. Click on Auto. This method might
be considered "subjective cheating" in terms of pulse recognition,
but the measurements are taken by Sound Ruler objectively, within the area that
you delimited. Notice that this method can also be used to reduce the number
of pulses. If Sound Ruler recognizes one pulse as being two, you can click on
each side of the pair, hit "Auto" and they will be recognized as one
pulse.
Mouse
Logging
1) Go to the menu "File"-"Open",
and open the example file H_versicolor.wav from the <soundruler>\sounds
directory (see step 1 of the example " easy calls").
2) Enter 4 in the box that asks about the length
that each section should have (see step 2 of the example " easy
calls").
3) Do some Manual or Auto call recognition (see
step 5 of the example " easy calls" or step 6
- Sample of the example " filtering").
4) Go to the menu "Tools"-"Mouse
Log"-"Start".
5) The TextPad opens with some header info. You
can add your notes to it.
6) Right-click on any of the four plots.
Sound Ruler will log the coordinates and the file info for each click on the
TextPad.
7) Go to the menu Tools - Mouse Log - Stop when
you are done.
GRAPHING
Exporting
graphs
1) Go to the menu "File"-"Open",
and open any sound file from the <soundruler>\sounds directory (see step
1 of the example " easy calls").
2)
Enter 4 in the box that asks about the length that each section should have
(see step 2 of the example " easy calls").
3) Do some Manual or Auto call recognition (see
step 5 of the example " easy calls" or step 6
- Sample of the example " filtering").
4) To save the current figure, use "File"-"Save
Figure" or "Edit"-"Copy Figure".
5) To export and edit clean figures, follow the
instruction bellow.
6) Go to the menu "File", "Export
Graphs". Choose which plots you want in your figure.
7) A new figure opens, with the plots you requested
on it, in a default arrangement.
8) The Export window is still open. Go back to
it and use the Edit Graphs controls to edit, move or resize the plots inside
the figure. You can resize the figure itself by dragging its edges with the
mouse.
9) Now go to the menu "Edit"-"Copy
Figure" or "File"-"Save Figure" to save the graphs.
Making
batch series of customized figures
1) Repeat the procedure described above (steps
of 1-8 of example "exporting graphs").
2) When you have the figure and the plots arranged
the way you want:
a- Minimize your
new figure window (do not close it).
b- Return to Sound
Ruler's main window and move to the next call.
c- Now expand your
figure window again and go to the menu RefreshData.
3) The data for the new call replaces the data for the old call inside your customized graph. Iterate through this process saving or copying in between, and you can quickly generate series of standardized custom figures for many signals.
4)
Save your customized figure in the .fig format, and you will be able to open
it later, refresh the data and do further editing. This way you can generate
your library of template files containing the formats that you use most frequently
for quickly producing series of standardized graphs.
Back
to top | Home
TOOLS
Calibrating
amplitude
1) Go to the menu "File"-"Open",
and open the example file amp_calib.wav from the <soundruler>\sounds directory
(see step 1 of the example " easy calls").
2) Enter 4 in the box that asks about the length
that each section should have (see step 2 of the example " easy
calls").
3) Click twice on the Section plot to select a
section of about 0.5 s of the calibration beep. Do not select the edges of the
sound.
4) Go to the menu "Tools"-"Amplitude
Calibration". Enter 90 and choose dB from the list. Click OK.
5) The Section is plotted again, but now the y
axis shows the amplitude in the calibrated scale.
NOTE: The measurements and the oscilograms are
calibrated, but for now, the spectrogram and power spectrum data are not affected
by amplitude calibration.
6) Go to the menu "File"-"Open",
open and analyze any sound file from the <soundruler>\sounds directory.
The oscillogram plots and measurements will be calibrated.
Building
a tuning curve filter
1) Go to the menu "File"-"Open",
and open the example file noise.wav from the <soundruler>\sounds directory
(see step 1 of the example " easy calls").
These are synthetic calls built with white noise,
which is characterized by random distribution of energy across all frequencies.
Select a call and perform a manual call recogntion. Notice how homogeneously
gray the spectrogram is and how flat the power spectrum is.
2) Go to the menu "Tools"-"Tuning Curve" and click on the Build Filter button.
3) Open the example filter file P_pustulosus.txt
from the <soundruler>\filters directory.
4)
A plot is generated showing the data that specify the filter (in blue) and the
actual profile of the filter that was generated (in black).
5) Change the Filter Order field to play with
the tradeoff between the complexity of the filter function and how well it matches
the data. Numerical errors might occur at very high filter orders.
6) Click on the Apply Filter button. The Section
Oscillogram will be updated with the filtered signal. Notice how the power spectrum
and spectrogram of the remaining sound match the profile of the filter.
7)
Try applying the P_pustulosus.txt filter to the sound file P_pustulosus.wav
(open the file from the <soundruler>\sounds directory) and listen to the
sound modified by the filter.
Copy and edit the example filter files to create your
own filters.
NOTE:
The Tuning Curve filter is independent from the filters specified on the Spectrogram.
Cross
Correlation Analysis
1) Go to the menu "File"-"Open",
and open the example file Psteckeri.wav from the <soundruler>\sounds directory
(see step 1 of the example " easy calls").
2) Enter 4 in the box that asks about the length
that each section should have (see step 2 of the example " easy
calls").
3) Do some Manual or Auto call recognition (see
step 5 of the example " easy calls" or step 6
- Sample of the example " filtering").
4) After a call is plotted click on the CCA button
by the Calls list to assign it for cross correlation analysis. A confirmation
text appears in the status bar on the right bottom.
5) When you have selected your calls, go to "Tools"-"Cross Correlation" and select all three types of correlations. In addition, select Normalize Amplitude.
6) Click on the Run button. Give it some file name to save a spreadsheet version of the output. The TextPad will also open for immediate inspection of the results.
7) To get an idea of the precision of this method, repeat the analysis with the files cca1.wav, cca2.wav and cca3.wav from the <soundruler>\sounds directory. This are synthetic sounds that start at 1, 2, or 3 seconds, have a duration of 0.4 seconds and frequencies of 1, 2, or 4 kHz. Changes in frequency and amplitude happen at about 0.1 s within the call, so you will be looking at delays that are multiples of 1000 Hz and 100 ms. Notice that the first sound in cca3.wav has a gradual onset, that will sligthly affect both the correlations and the delays.
Problem
solving with Sound Math
1) Go to the menu "Tools"-"Sound
Math".
2)
Change the value of any field and watch the effect on related fields.
3) Click on the Copy button at any time to export
the current values to the TexPad.