Why is a dictaphone the default way to interact with speech-to-text software in radiology departments? I am truly flummoxed. Every reading room I walk into looks like this:
An entire hand is used to do nothing other than hold the dictaphone to one’s mouth for most of the day. This hand could be used for many other useful things — answering phone calls, hitting keyboard shortcuts to speed up one’s workflow, scrolling through clinical history, sipping coffee to stay vigilant.
A simple stand could equally well serve this essential function of elevating the microphone to sufficient proximity to one’s mouth to be useful. As could, say, a microphone mounted directly to one’s face — a ‘head set’, of sorts. A cursory review of Google Images suggests that the rest of the world has moved on to this technology … perhaps we should, too?
I’ve started using an setup in my own practice that costs around $300 to put together. Several colleagues have asked me how I put it together, so I am sharing the details below. I make no claim that this is optimal — just that this is what I am using.
A headset
The first thing we need is, yes, a headset. My first instinct was to get a nice standalone Bluetooth headset and pair it with one of the many $10 bluetooth USB dongles on Amazon, something like this.
These USB bluetooth dongles worked on a plug-and-play basis on a Windows 10 computer I had sitting around at home. However, none of the 3 different ones I tried worked on any of the Windows 10 workstations in the reading rooms. I suspect that lacking admin rights on these workstations foiled the plug-and-play driver installation process. I could have asked IT to install it on a given workstation, but this wouldn’t have done much for me, since I am always rotating around to different computers and different reading rooms. It’s not practical for anything used by a radiology resident to have to be manually installed on a given computer, although this may work for an attending who consistently uses the same workstation.
I initially gave up on wireless and used a $40 wired Sennheiser USB headset that worked out of the box on these workstations and did not require admin rights for driver installation. The speech recognition worked at least as well for me as the $300 Phillips SpeechMike I was using previously, and it was much better ergonomically than the dictaphone.
But the cord was annoying, especially having to take it on and off whenever I went to someone else’s workstation. I kept on searching and found the Plantronics Voyager 5200 UC headset for $150:
This is what I’m using now, and is basically ideal for my purposes. It comes with a wireless USB dongle that works on the PACS workstations as pure plug-and-play without any requiring any new drivers (the computer seems to see it as an accessory sound card rather than a Bluetooth accessory). It comes with a handy charging/carrying case that can provide an extra full charge when needed. It’s small, unobtrusive, and comfortable to wear the whole day. The microphone is pretty sensitive — I think the cheaper, larger, wired Sennheiser headset was actually a little more sensitive, but the wirelessness is more than worth the slight trade-off.
Update 11/2020: I still use the Plantronics Voyager 5200 on a day to day basis. The charge lasts ~6 hours during continuous use — so if you’re a resident with noon conference when you can leave it plugged in to recharge you can easily make it through the day. But when I am on call overnight, I find that I am always reaching for the Sennheiser. Its sensitivity is incredible, and when I am mumbling quickly and have minimal tolerance/time for correcting mis-transcription, it delivers — and in my opinion markedly outperforms the more expensive Phillips dictaphone).
A dictate button
The first question I always get after I’ve explained the headset is, “but how do you dictate?” And it is true, the headset lacks the big red dictate button that the SpeechMike conveniently featured.
The solution is to map another button to the ‘dictate’ button in your speech recognition software. Most speech recognition software will let you map a dedicated keyboard key to this function. On my Fluency profile, I have ‘F1’ mapped to this function.
Option 1: Press F1 on the keyboard every time you want to start or stop dictating. I didn’t really consider this, since it would keep my left hand hovering over F1 throughout the day, barely an improvement on the dictaphone.
Option 2: You can get a special gaming mouse that maps mouse buttons to user-defined keys and stores those internally. I.e., you program the ‘trigger’ button on the mouse to be ‘F1’, and then whenever you plug that mouse into a new computer and press that button, it sends F1. The key here is that the mapping is stored in the mouse itself, so that no software needs to be installed on the computer on which the mouse is used. I initially used the wired Logitech G502 Proteus mouse and then moved onto the $150 wireless Logitech G502 Lightspeed:
I do really like this mouse — it’s the one I use day to day. The amount that a mouse can improve one’s life is pretty tightly upper bounded, but this feels great and the trigger button is an excellent candidate to replace the dictate button on your typical dictaphone. It has a bunch of extra buttons you can map to useful things like forward/backward tab, delete, etc. to speed up your dictations.
Also, use a mousepad! They help a lot.
To program the buttons, you have to install a bloated software package called Logitech Gaming Software (for the wired G502 Proteus) or Logitech G-Hub (for the wireless G502 Lightspeed) onto any computer and program the buttons. You can uninstall it after you’ve programmed the buttons if you hate the software (I did).
Option 3: A lot of people like to use the wireless Logitech productivity mice like the MX Master. I modestly reiterate that I, personally, strongly prefer the gaming mouse above, and think that the exceptional design is worth paying a premium for. But if you are committed to your MX Master:
The wireless productivity mice have extra buttons you can map to user-defined keystrokes. However, unlike the Logitech gaming mouse above, these mice do not store the key mappings internally. If you always use the same computer, you can download the Logitech Options software, ask IT to install it, and set up the mappings.
If you don’t, I recommend using a program called AutoHotkey. This program is a great productivity hack for people who spend a lot of their days making arbitrary keystrokes/mouseclicks (…radiologists!). This requires more technical comfort to use effectively, as it’s a scripting language and you will have to do some customization to make it work. I’m assuming you have a Windows-based PACS workstation. You can create a simple script to map a mouse button a key of your choice as follows:
1. Download and run AutoHotkey (AHK).
You can download the Windows executable. It does not require admin privileges to run. AutoHotkey scripts end with “.ahk.”
2. Download and run the KeypressOSD script.
You can confirm it’s running by looking at the bottom right of the screen — you should see a little green square with a white Comic Sans H in it. If you hover over, it should say “KeypressOSD.ahk”. This means that this script is loaded and waiting to act.
3. Press the mouse button you want to map to ‘start/stop dictation’ on your mouse (it’ll work for keyboard presses too). You’ll see its AutoHotkey name show up on the screen. For example, on an MX Master 2 I have, one of the extra buttons shows up as XButton1. Let’s say I wanted to map this to F1, which is my dictate function in Fluency. On a Windows machine, I’d open up Notepad, copy and paste the below into it, and save it as “keymap.ahk” (with surrounding quotes so Notepad doesn’t append .txt to the end of the filename):
XButton1::
Send,{F1}
return
What do these few lines do? They map what the computer sees as XButton1 to press the F1 key — ie, whenever you press this mouse button, the text to speech software will see F1, and will start or stop dictation.
If you go this route, you’ll need to launch this script as part of your (inevitably already cumbersome) stack before starting to work.
Other notes
I got one these USB 4-in-1 things so I could leave both the wireless mouse and headset USB dongles plugged into it and have just 1 thing to plug into the workstation each day.
And — for the enterprising— you can use AutoHotkey for way more than mapping keys. It’s an incredible productivity tool for people forced to use clunky software that frequently requires a Byzantine series of mouseclicks / keystrokes (us!). I use it for a bunch of stuff, including automatic scrolling to reduce strain and to automate elaborate series of repeated mouse clicks, but it does require comfort with scripting to use effectively.
Maybe I’ll write more about AHK another time. For now, I’d be delighted if we could just agree to ditch the dictaphone.
