Telemed MicrUs documentation
Hardware, experimental setup
UCLA Phonetics Lab
Table of contents
Instructions for using the lab's ultrasound facilities are split into two parts. This page covers the logistics of what you need to set up to collect data or run in-class demos of speech articulation as well as the human interaction element: how to adjust the stabilization headset and probe to maximize participant comfort while ensuring a good view into the oral cavity.
Our hardware consists of the Telemed MicrUs, the Telemed MC4 or MC10 convex probes, a USB audio interface, and a few connector cables. This page also serves as a guide to the accompanying software Echo Wave II, a license for which is required for full functionality of the MicrUs, and gives a brief overview of how to change settings in the device using the software.
Collecting data or running a demo requires a suitable work station. One data acquisition laptop is currently configured to communicate with the ultrasound device. If you would like another device to be configured for collecting ultrasound data, please contact Matt or Henry. If you plan to collect ultrasound data or use the device for an in-class demo, be sure to reserve the ultrasound device and laptop in addition to whatever lab room you need to book.
Demos, and some configuration before data collection, are done in Echo Wave II, which is proprietary software provided by Telemed, the probe manufacturer. This is only installed on the data collection laptop at the moment.
There are two ways to collect ultrasound data. The Articulate Assistant Advanced (AAA) software provided with the ultrasound device comes with its own tutorials. There are numerous data processing options that are available for data collected and stored using AAA. However, the data collection process itself is very streamlined and is mainly designed with SLP clinicians in mind, so it may not work for all experimental designs.
The lab's Python utilities are a good deal more flexible since they are based on the OpenSesame experiment GUI. OpenSesame is free and open-source, and it has the advantage of being written in Python (which is also free), which means we can easily define custom code blocks for communicating with the ultrasound device based on various triggers. A full tutorial for OpenSesame is beyond the scope of this documentation, but it has a relatively large user base and a lot of useful online discussion of common issues, as well as extensive documentation.
Several template OpenSesame experiments can be found here. In OpenSesame, you can customize these script templates to suit your own needs, either using the GUI or by directly editing the Python code in various sections. Unfortunately, you will only be able to run the script on a Windows computer which the ultrasound probes are connected to.
To record the ultrasound signal and align it to audio, an exchange of information between the ultrasound device, a USB audio device, and the computer needs to take place. Once you've found a suitable workstation, you'll need to set up your hardware and start the communication between the ultrasound device and the workstation.
For this process you will need these supplies, which are all stored in the camera case.
- Micro unit (small, flat silver box)
- PStretch unit (stubby silver box with black bumpers
- Scarlett 2i2 USB audio interface
- Three unattached cables: copper-colored and long; red; silver
- One ultrasound probe of your choice
Hardware setup should take place before headset fitting, since the speaker's presence is not needed and arranging all the materials can be somewhat time consuming. This is documented on the other page and has its own material requirements.
Connect the Focusrite Scarlett 2i2 ("Scarlett") to your computer using the red USB cable (port is on back of Scarlett).
Connect your microphone to Scarlett using the XLR port (see picture below).
Turn on the Scarlett's phantom power by pressing the button next to the XLR port (see picture above). It will light up red when on.
Connect the PStretch unit (small box with black "bumpers" on either end; see picture below) to the Scarlett's 1/4" headphone jack using the 1/4" headphone to BNC cable. The BNC end of this cable must be attached to the PStretch's SYNC OUT port (this is labeled on the back of the device; on the left in the image below).
Attach the BNC to mini BNC cable (copper-colored and more like a wire) to the PStretch's SYNC IN port. The larger end of the cable attaches to the PStretch; you will need to twist the cable about 90 degrees clockwise to lock it in into place. The cable we use nowadays is longer than the one in the pictures here.
Attach the other end of the BNC to mini BNC cable to the Micro's "Frame Strobe" mini BNC port (pictured below). There is a small hex nut attached to the mini BNC end that you have to tighten by hand.
Double-check that all connections to and from the PStretch are to and from the correct ports, as shown below, and all connections are fully "clicked" in. This is important! Most problems with data collection are due to bad connections to or from the PStretch, or due to a loose connection in one of the headphone jacks. If something doesn't look right, review steps 5-7 above.
Finish making the connections to the laptop: plug the PStretch into another USB port on the computer using the black USB cable that is attached to the unit. Using the silver mini USB to USB cable, connect the Micro's mini USB port to another USB port on your computer. Note that you should use the "first" USB connection on the cable (the one which also has another USB plug coming off of it), which is pointing towards the top of the image below.
Finally, take your desired probe out of storage and attach it to the multi-pin socket on the front side of the Micro.
The imaging settings are adjusted in Echo Wave II. We will generally load preset combinations of settings into Echo Wave II. This is ideally done before you fit the stabilization headset and probe to your speaker.
To confirm your settings, and then change them as needed:
Run Echo Wave II with the desired probe connected. This often takes a few seconds to start up.
Check the settings. Echo Wave II automatically uses the settings that were used in the last session for that probe, whether or not they were a preset. Four presets are available, two for each of the two probes. These are available as .dat files stored on the data collection laptop, and which can also be downloaded here.
Normal setting: records at 82 frames per second; full field of view. Good for most speech sounds and study questions.
Large probe (Telemed MC4): telemed-MC4.dat
Small probe (Telemed MC10): telemed-MC10.dat
Faster setting: about 120 frames per second; smaller field of view (60 percent of normal). Best used for flaps, taps, and other similarly quick (sub)segments, or where especially fine temporal detail is needed for other reasons.
Large probe (Telemed MC4): telemed-MC4-60pct.dat
Small probe (Telemed MC10): telemed-MC10-60pct.dat
Depending on your needs, leave the existing preset in place or load a new preset by importing a .dat file. Click the small fan icon in the upper left of the screen (near the probe name in the upper left corner; see image above). The fan icon will produce a drop-down menu when clicked; click "Presets", then "Apply Settings from File" at bottom left.
If needed, you can also create custom imaging settings and save them as a new .dat file. However, the presets generally work well and you will probably not need to reinvent the wheel here.
With Echo Wave II open to check the probe position, you will now want to put the stabilization helmet on your speaker, then put the probe into position in the stabilization headset. This process is explained in more detail here.
After you are satisfied with your probe placement, you must close Echo Wave II to collect data using the experiment software. Whatever settings you were using here are the ones that will be used during recording, even though Echo Wave II is not open.
Software 2: the experiment script
To start your OpenSesame session, you will need to do the following in order. Before you start you should turn off the device's wifi (which makes the connection to the ultrasound more stable), and make sure that the computer is not scheduled to restart for software updates.
Open the Anaconda Prompt and type in the following: C:\Users\linguist\Anaconda2\Scripts\opensesame.exe. This will open OpenSesame with the required permissions. Using this command in another terminal will not work.
A blank experiment loads by default. Open your experiment file of choice. We keep several template experiments in the Box folder here; these can be adapted for your needs.
Before running the experiment, open the Windows PowerShell and type in: C:\echobserv.200.exe, which starts up the communication between the ultrasound device and the collection laptop's hard drive.
At this point, run your experiment. A small MATLAB console will appear, showing a live video of the ultrasound signal. You cannot close this window, since it will cut off communication with the ultrasound. Since you will need to use this to monitor the swallow task and bite plate task, we suggest clicking on the double, small green arrow (Run in Window) instead of the large green arrow (Run Full-screen). This will open up a moveable window instead of full-screening the experiment. Once the landmark tasks are done, you should cover the MATLAB window up with the experiment window so that the participant can't see their own articulations.
The experiment template contains code which displays "slides" to the participant for three types of events. You will generally want these to happen in this order:
Your experiment will probably begin with the swallow task and bite plate task, which we use to get more details on the recording session's physical frame of reference. These each have their own script items which can be repeated a desired number of times. Essentially, each triggers an ultrasound recording that lasts from a starting keypress to an ending keypress, allowing the participant to take as long as needed.
Slides familiarizing the participant with the study usually occur after the swallow and bite plate tasks. These advance as the participant clicks through.
Finally, most of the runtime is taken up by a loop which runs experiment blocks a specified number of times. Trials can be randomized. These are again self-paced, with the recording restarting whenever the participant clicks through to indicate they have finished reading the presented item.
We suggest providing a break every 10-20 trials, between blocks. To avoid fatiguing participants, we also don't suggest trying to record more than 500-600 sentence-length trials in total. This takes about 30 minutes depending on the participant's speed, and spending much longer than that in the headset can be fatiguing.
Displaying a live feed of someone's tongue with the ultrasound device is much simpler than doing data collection. The Echo Wave II software is designed for just this task in clinical or diagnostic settings, and the probe usually doesn't need to be stabilized. If Echo Wave II is opened full-screen and projected at the front of the room or shared over Zoom, students can usually make out all the relevant details of lingual articulation.
- Connect the ultrasound device (the small, silver box) to the laptop using the silver mini-USB to USB cable. You should use the "first" USB plug rather than the second one branching off it. See picture below.
- Select the probe you would like to use and connect it to the multi-pin socket on the other end of the ultrasound device.
- Double-click the Echo Wave II 2.6.2 shortcut on the desktop and allow it to run. Start-up takes a few seconds.
From here, follow the instructions for orienting the probe on yourself or a volunteer speaker, minus the stabilization headset (which is not usually needed for simple demos).
Ultrasound is useful for visualizing lingual articulation but cannot visualize much else. Since the probe will probably not be stabilized, height or degree of constriction will also be hard to eyeball. The most useful demos focus on big, visible articulations, which include:
Showing what various complex lingual segments look like, such as flaps, taps, /l/, /ɹ/, or other liquids that volunteer speakers might produce. Vibration of the tongue tip in the trill [r] is also often visible in the signal.
If you are demoing for Ling 103, ultrasound is great for troubleshooting students' production of clicks, as well as retroflex, palatal, uvular, and pharyngeal places of articulation. You can produce these yourself to demonstrate or get a native speaker of a language which has these contrasts (Arabic, Tamil, etc) to volunteer.
Connected speech is generally fun to watch. Try simply recording yourself while explaining what's going on at a normal speaking rate.
The ultrasound will run continuously unless you hit the "freeze" button, which is in the lower left corner of the screen. Freeze the signal if you are not recording something in motion, or if you need a snapshot of a particular tongue configuration.