-
Notifications
You must be signed in to change notification settings - Fork 0
[ AAA EMA & Vicon User Manual PDF from 2014 ]
![]() |
|---|
Copyright 2003-2014 Articulate Instruments Ltd^1
Version 2.
Address:
Articulate Instruments Limited
Queen Margaret Campus, Queen Margaret University Drive
Musselburgh EH21 6UU
UK
Phone:0131 474 0000
Fax:0131 4740001
Email:support@articulateinstruments.com
Copyright 2003-2014 Articulate Instruments Ltd^2
Information in this document is subject to change without notice and does not represent a commitment on the part of Articulate Instruments Ltd. The software described in this document is furnished under a license agreement and may be used and copied only in accordance with the terms of the license agreement. It is against the law to copy the software on any other medium except as specifically allowed in the license agreement. No part of this manual may be published, reproduced or transmitted in any form or by any means, electronic or mechanical, including photocopying and recording, for any purpose without the written permission of Articulate Instruments Ltd.
- Copyright 2003-2014 Articulate Instruments Ltd
- Importing EMA data
- TYPE 1 Importing EMA and associated wav data
- Converting raw 3D data to x/y/z analysis values.........................................
- TYPE 2 Importing EMA into existing recordings in AAA
- TYPE 3 Importing EMA data after AAA was used to record the audio........
- More about analysis values
- Plotting EMA analysis values....................................................................
- Recording VICON Data
- Importing VICON Data
- EMA Analysis...............................................................................................
- The 3D EMA Display
- EMA Analysis Values................................................................................
- The Maths analysis value
- Smoothing..............................................................................................
- Find peaks and valleys
- Finding a region 10% either side of a peak............................................
- EMA Analysis Values................................................................................
- Shortcut Keys
- Index............................................................................................................
Copyright 2003-2014 Articulate Instruments Ltd^4
Importing can be done in four ways
- By importing EMA AG50x pos files and associated wav files together plus PRAAT textgrid files if present.
- By importing EMA AG50x pos files and associating manually each pos file with a wav file already present in the current project.
- Importing EMA AG50x pos files to match with wav data originally recorded at the same time with the multichannel system (This differs from 2 above because the recordings in the project retain information about which pos files match them and so the matching is done automatically)
- Importing AG200 *.0?? , .1?? , *.2?? , *.M?? data files and PRAAT textgrid files if available.
To import EMA and audio data that you have recorded using the standard Carstens equipment:
- Load or create a project that you want the EMA and wav data to be imported into.
- Select the ‘File|Import EMA...’ menu option to open the EMA import dialogue.
- Select the “Import into new recordings” tab.
Figure 1 EMA Import Dialogue with channel name editor open and the names "Right Jaw" and "Left Jaw" added
- Select the folder containing the EMA pos files and associated wav files. Note: The wav files must be in the same folder as the pos files and have filenames that match.
Copyright 2003-2014 Articulate Instruments Ltd^5
- Specify the channels name assignments. (If appropriate names don’t appear in the dropdown list then add them to the list using the Edit Channel Names button).
- Check the boxes corresponding to the recordings you wish to import with the specified channel assignments (Figure 2). Note that this allows the assignment to be changed part way through a session.
- Click OK to import the data into the current Note: Re-importing into the same client will add new identically named recordings.
Figure 2 EMA Import Dialogue with recordings selected and Channel names assigned.
AAA cannot display or perform calculations on raw 3D data. The raw data first needs to be converted to 3 displacement values. Once the raw data is imported a dialogue (Figure 3) will automatically appear with the names of all the channels. Raw 3D data cannot be viewed inside AAA. It must be converted to analysis values. For eachselected channel name, three analysis values will be created with x, y and z displacement vales respectively. In most cases, all of the channel names should be selected and converted. The only reason for deselecting (not converting) a channel name is if that channel is never going to be required for analysis.
Note: The raw data also contains information on the orientation of the
sensor but this data is not currently offered as an analysis value within AAA.
If you need this, contact Articulate Instruments Ltd.
Copyright 2003-2014 Articulate Instruments Ltd^6
Figure 3 Create analysis values from raw 3D data
- Data should then appear in the Prompt list window.
For completeness it is possible to import EMA data recorded using Carstens equipment into an existing client with existing wav recordings:
This option is unlikely to be required but is included for historical reasons.
- Load or create a project that you want the EMA data to be imported into.
Copyright 2003-2014 Articulate Instruments Ltd^7
- Select the ‘File|Import EMA...’ menu option to open the EMA import dialogue.
- Select the “Import into existing recordings” tab
- Browse to find the folder with the EMA pos data and check equal numbers of pos files and recordings.
- Click OK Note: the OK button will be disabled unless equal numbers of pos and wav files are checked. If multiple files are selected, there is currently no way to select how they match up. Contact us if you need this.
Figure 4 Importing EMA data into existing recordings
If AAA was used to record EMA and audio then the audio will already be stored in AAA along with the name of the *.pos file that is associated with it. You only need to select the processed pos files ( ensure that the original names are retained after post-processing ) and the import function will automatically import them into the correct recording.
- Load or create a project that you want the EMA data to be imported into.
- Select the ‘File|Import EMA...’ menu option to open the EMA import dialogue.
- Select the “Import into existing recordings” tab
- Browse to find the folder with the EMA pos data and check the pos files that are to be imported.
- Click OK. Note: Re-importing will overwrite the EMA data. Note: Pos file name stems must match the original *.amp file name stems. ie. if the sweep recorded files 0001-0010.amp then after post-processing the the pos piles should be 0001-0010.pos
Copyright 2003-2014 Articulate Instruments Ltd^8
It is also possible to convert raw data to analysis values at any time after the import using the Analysis Values dialogue. To do this:
- Click on the Menu option ‘Options | Analysis Values...’
- Click on the button and the “Add EMA Analysis” box will pop up (Figure 5).
- Select a channel name from the list that corresponds to a channel you have just imported. Note that if analysis values with the same name have been set up for previous recordings, they do not need to be set up again. All three x,y and z displacement values are automatically created when the OK button is clicked.
Figure 5 Analysis Values Dialogue with Add 3D Data (EMA/VICON) Analysis option selected
Copyright 2003-2014 Articulate Instruments Ltd^9
Figure 6 Analysis values dialogue with 3D Data (EMA/VICON) xyz displacement values loaded for each channel.
The Display tab provides options to hide analysis values in the Analysis Values Text window, in the Edit charts list and in the list of values to plot in the 3D plot window. This is useful to de-clutter these lists. For instance, if only the y-values are of interest then the x and z values can be hidden, reducing the length of the list that needs to be scrolled through to find the value of interest.
Copyright 2003-2014 Articulate Instruments Ltd^10
Figure 7 Analysis values dialogue showing the Display tab which provides options to hide the value from being displayed. Useful for declutterring lists.
Now that the raw data has been converted into analysis values it is possible to plot them and to further process the displacement values to smooth or differentiate. Notice that all the EMA displacement values are automatically selected for display in the current analysis chart. See the section on Editing analysis charts in the AAA User Manual for help in rearranging the display.
As well as displaying EMA values in a chart with magnitude vs. time, it is possible to plot the values in 3 spatial dimensions in the ‘Plot window’. This can be done in the following way:
- Open the "Analyse" Task Window. You should see all of the 3D values charted in the "Analysis chart" window.
Copyright 2003-2014 Articulate Instruments Ltd^11
Figure 8 Analysis Task Window showing all the 3D values in the Analysis Chart window and an empty plot window.
- Right click on the plot window and select the option from the popup menu to open the Edit Plot dialogue.
- Select the pane axes to be 3D and click to add one channel. This will open the dialogue showing all the analysis values available in the current project. ( Figure 9)
Plot window
Analysis Chart
Window
Copyright 2003-2014 Articulate Instruments Ltd^12
Figure 9 Select the 3D Analysis values to be plotted
- All of the analysis values will be loaded and can be accessed by clicking the tabs marked 1 to N
- Select the ‘Plotting’ tab (Figure 11) and set the "Fade with age"
option to show the trace of the path of the sensor.
- Set the symbol to represent each 3D value.
- Tongue sensors or other groups of sensors can be connected with lines by clicking on the graph number to be linked to the currently viewed graph e.g. if you want to connect graphed point 4 to the current point (7) click on 4 then select the colour and width of the connecting line.
Copyright 2003-2014 Articulate Instruments Ltd^13
Figure 10 Edit Plot Dialogue with selected 3D channels loaded as tabs (10 values shown here).
Figure 11 Edit Plot dialogue plotting parameters
Copyright 2003-2014 Articulate Instruments Ltd^14
Figure 12 Edit Plot dialogue with "Use common axis selected". Use this to set each axis to be the same for every channel.
Figure 13 Edit Plot Dialogue showing "Plotting" tab. By default all of the 3D values are shown as "+" symbols.
- Optionally, it is possible to set up the Head Drawing parameters using
the button.
Copyright 2003-2014 Articulate Instruments Ltd^15
Figure 14 Head Drawing Setup Dialogue
a. Click and drag the highlighted marker labelled ‘First fixed point’
to the Bridge of the nose on the outline of the head and click on
the analysis value named Bridge of nose. (see Figure 14).
b. Click and drag the highlighted marker labelled ‘Second fixed
point’ to the upper incisor position on the outline of the head (or
wherever the second reference point is) and select the
appropriate analysis value from the list.
c. Click and drag the highlighted marker labelled ‘Point defines
plane of head (can be mean of two points)’, move the marker
and select left ear AND right ear in the list
d. Select the final marker and move to where the jaw sensor is
and select jaw from list.
Copyright 2003-2014 Articulate Instruments Ltd^16
Figure 15 Head drawing setup with markers in typical locations
Figure 16 Selecting sensors to be linked together
Copyright 2003-2014 Articulate Instruments Ltd^17
Figure 17 Plot Window. Resulting 3D plot with tongue coils connected and silhouette of 2D head visible
EMA Analysis...............................................................................................
The previous section showed how to prepare EMA data for display in a 3D. This 3D plot is promarily designed to provide visual confirmation of the integrity of the data, to confirm that the channel assignment is correct and that sensors are functioning as expected.
Smoothing..............................................................................................
Smoothing of EMA displacement data is important as it allows peak detecting algorithms to work and is crucial in removing the effect of noise in the measurement when calculating velocity and acceleration traces. Low- pass digital FIR filters can be used to remove high frequency components but if the low-pass cutoff frequency is set too low then some of the components of the underlying displacement trace will be lost. In AAA we use
Copyright 2003-2014 Articulate Instruments Ltd^18
moving average smoothing filters and in particular the Savitzky-Golay smoothing filter. The Savitzky Golay filter is a generalised moving average where the filter coefficients are derived by performing an unweighted linear least squares fit using a polynomial of the specified degree (2 or 4). As a rough guideline, best results are obtained when the full width of the degree 4 Savitzky-Golay filter is between 1 and 2 times the Full Width at Half Maximum of the desired features in the data. These values can all be displayed in the ‘Analysis Values’ chart display (Figure 18).
Figure 18 Raw Tangential velocity (blue) ; Smoothed Tangential Velocity (black) using rectangular window of 20ms
Figure 19 Raw Tangential velocity (blue) ; Smoothed Tangential Velocity (black) using triangular window of 20ms
Copyright 2003-2014 Articulate Instruments Ltd^19
Figure 20 Raw Tangential velocity (blue) ; Smoothed Tangential Velocity (black) using 2nd order Savitzky-Golay window of 20ms
When smoothing EMA displacement data, a good default setting is to calculate the smoothed data every 1ms with a 20ms window using the 2 nd order Savitsky Golay (savgol) weighting function. This results in smooth data that fits the underlying contour of the data very well. Figure 20 shows the improved results compared to using a moving average smoother with a rectangular or triangular window. Savitzky-Golay filters are often preferred because, when they are appropriately designed to match the waveform ofan oversampled signal corrupted by noise, they tend to preserve the width and height of peaks in the signal waveform. While such performance features are often explained in terms of the implicit polynomial fitting process (where it is assumed that the fitted polynomial slopes are matched to those of the signal) the reason for this behavior is also obvious from the frequency domain properties of the filters. Specifically, they have extremely flat passbands with modest attenuation in their stopbands. Furthermore, the symmetric S-G filters have zero phase so that features of the signal are not shifted. Thus, if the signal has most of its energy in the passband of the filter (implying significant over-sampling), the signal components are undistorted while some high-frequency noise is reduced but not completely eliminated. Of course, assuming that the signal is lowpass is equivalent to assuming that the signal is smooth enough to be represented by a polynomial of high enough degree. However, S-G filters are often used in situations where a direct frequency-domain specification is more precise or more easily related to models for signal production. Toward the end of quantifying the design of S-G filters, there is an empirical relationship in (1) between 3 dB frequency and the parameters M and N.
The nominal normalized cutoff (3 dB) frequency, fc = ωc/π, depends on both the implicit polynomial order N and the length of the impulse response, (2M +
Copyright 2003-2014 Articulate Instruments Ltd^20
1). If M is fixed, the passband of the filter gets wider approximately in proportion to N. The cutoff frequency depends inversely on M.
An approximate empirical relation for fc as a function of N and M. In AAA, N=2 or 4 2M+1 defaults to 20ms at 200Hz = 40 so M~=
(1) fc N . M – .
at
ms Hz M
Hz for order and Hz for order
at ms Hz M Hz for order
at ms Hz M Hz for order
We recommend order 4 and a window in the range 25-40ms (assuming 200Hz EMA). The wider the window the smoother the data but it may smooth away high velocity movements for taps and flaps, and /l/
The “Find” function can be invoked by right clicking on the analysis values chart window. The “Finding Frames of Maximum and Minimum Contact “ section of the AAA manual describes the components of the “Find” dialogue and provides an example of a function to find EPG closure regions. This section will describe functions useful in the analysis of EMA data.
This can be applied to displacement or tangential velocity traces for a given EMA sensor.
Stage 1 Find peak in EMA Tongue Blade Tangential Velocity (TB2 Tang Vel) within 50ms either side of where the mouse is clicked
Copyright 2003-2014 Articulate Instruments Ltd^21
Stage 2 Find last valley before the peak Search a region 10 seconds before the peak and find the last valley in this region.
Mouse
clicked here
Finds Peak
(Pk) here
+- 50ms
search area
Change these to –20 and
20ms for a tighter peak
search
Change this to “Sel” if you
want to search for the peak
within a manually selected
region
Copyright 2003-2014 Articulate Instruments Ltd^22
Stage 3 Find the “Onset” point 10% of the way from the valley to the peak Define a region starting at the valley and ending at the peak set the variable V to the value
Last minimum before
peak
Finds onset
point: 10%
(min to Pk)
re
Copyright 2003-2014 Articulate Instruments Ltd^23
Stage 4 Repeat stages 1-3 to find 10% “Offset” point up from the first valley after the peak.
Change this % value to
your preference
Copyright 2003-2014 Articulate Instruments Ltd^24
Stage 5 Annotate the regions onset to peak, peak to offset and onset to offset.
Copyright 2003-2014 Articulate Instruments Ltd^25
Figure 21 Result of executing the “Percent of peak region” Find script
Help F1 Playback F2
Copyright 2003-2014 Articulate Instruments Ltd^26
Record F3 Select visible region of waveform Ctrl+A Adjust selected region/ selected annotation Ctrl + drag mouse Select region in Palates or Analysis values windows Shift + drag mouse
Revision Date Notes 2.11 19/05/11 Importing AG200 data instructions 2.12 01/06/11 Advice on smoothing filter settings 2.15 15/03/14 Revised import instructions updated after revision 215 of the AAA software.
Copyright 2003-2014 Articulate Instruments Ltd^27
Index............................................................................................................
Add EMA anaylsis........................ 8 Analysis Values ................... 17, 20 Analysis Values Display............. 25 Edit Plot ..................................... 11 EMA 3D Display......................... 17 Fade with age ............................ 12
Head Drawing setup .................. 14
Importing
Files.............................. 4, 5, 6, 7
Plotting tab ................................ 12
pos files ....................................... 4

