with Lindsey J. Bertomen
Voice recognition software: Report-writing without the writing
Two voice-to-text tools can transcribe as fast as you can talk
Since I have used Dragon NaturallySpeaking — software that turns voice into text three times faster than typing — for more than 12 years, I know the "ins and outs" of voice recognition software, and in my opinion, public safety would reap the highest benefit of all the industries that could use it.
I tested two Nuance products that I'm pretty sure will be absolutely profound to those who don't already use them: Dragon NaturallySpeaking Version 11.5 Premium Edition and MacSpeech Scribe, a personal transcription software package designed for Mac OS X.
Dragon NaturallySpeaking Version 11.5 Premium Edition
Most people think that speech recognition software is designed to recognize individual words and translate them into text, but the software actually recognizes speech patterns. In its simplest terms (and the only terms I can really understand), audio input is turned into patterns and the software chooses the most likely pattern based on the history of the speaker. The computer can do this quickly by comparing the input with what it already knows about the speaker.
Dragon NaturallySpeaking takes a few minutes for the software to learn the speech patterns of the user, which it stores as the user profile. Using my computer of average processor and speed, it took me about 15 minutes to set up my profile. The software provides selections of simple-to-read text that the user reads aloud. I read the Alice in Wonderland text selection by Lewis Carroll.
After about 15 minutes of training, the software could recognize speeds of more than 130 words per minute. I admit that I am an experienced user, but this is a good demonstration of how much better — and more accurate — Dragon NaturallySpeaking Version 11.5 is than my original purchase in the 1990s. Using the new version, I spent a single day dictating a little more than 10,000 words. I made one correction, and it was a proper noun that I had not trained the software to recognize.
The software trains constantly, improving its accuracy as it goes. Eventually, it becomes rather intimate with one's writing style. One can customize a word list, either by inputting it or correcting the spelling. For example, "RangeMaster” doesn't appear in the dictionary, but I use it often, so customizing it means I don't have to stop and ask for it to be written how I like.
Dragon NaturallySpeaking has easy-to-learn commands for correcting one's dictation. I use "scratch that” to tell the software to delete the last phrase of dictation — the hardest thing about dictating this article is the fact that I had to type “scratch that” rather than say it because it kept erasing the previous line!
I buy cheap computers. The one I am using to dictate this article is about the same or slower than most of the laptops I test for law enforcement. I'm telling you this because Dragon NaturallySpeaking Version 11.5 could be easily loaded onto the MDC's of an entire fleet of patrol cars and officers could dictate and proof their narratives in the car, making them more efficient.
Is it possible to use it in CAD? Depending on the system, yes. However, no one can type faster than a good dispatcher. No one can read people over the phone better than a good dispatcher either.
While on the road, I use an inexpensive Sony handheld digital recorder for my dictations, and similarly, officers could dictate their reports and then submit their recordings for transcription later. The premium version of Dragon NaturallySpeaking 11.5 is capable of taking a recording and transcribing it. This is one of the features that distinguishes the Premium edition. Because it supports multiple input devices using the same profile, it's not necessary to retrain it separately for each device.
Another feature of the premium edition is the capability of using the Bluetooth feature of an iPhone to make a wireless input device. This is pretty cool.
I used my digital recorder to record the previous paragraph. While it is preferable to use the software sitting at a computer, the digital recorder is the next best thing.
Like I said, I am an advanced user. As the software becomes trained, so does the user. Most people take a little while to get used to processing information while speaking it, which makes dictation difficult at first. When I first started using this kind of product, I would spend a few extra minutes writing a really good outline before speaking — which actually improved my report-writing skills.
Eventually the user will get to the point where they can simply speak their thoughts into the microphone, then edit later, which is excellent when putting together an incident report in chronological order.
Dragon NaturallySpeaking Premium Version 11.5 automatically recognizes common commands for software we use every day. This includes things like Microsoft Word, PowerPoint and Excel. It is very easy to maneuver throughout the documents and presentations using voice commands. I don't have to step up to my keyboard to save a document or title of presentation.
I can move my mouse anywhere on the screen without touching anything. There are commands to move it up or down (I’m not going to say them now) or one can use a "MouseGrid", which breaks the screen up into sectors where the user tells the mouse to go, based on the address of the grid. I generally use this feature, simply because it is very quick when one has several windows open at the same time. Yes, Dragon NaturallySpeaking will accommodate users who multitask, like me. I don't know everything about the software. Most users do the same thing I do: They have a set number of features that work for them and they use them well.
Nuance is the same company that has produced several really good products aimed at reducing distracted driving. The ones that work on my iPhone are free and I use them all. My favorite is Dragon Dictation for IPhone. This is a dictation app that converts text to speech which can be copied and pasted into anything. I can dictate onto my iPhone at least twice as fast as one can type. Let me rephrase that: as fast as a normal person can type. My daughter can easily outrun any dictation software.
Transcription software is important to the law enforcement community and I was also able to test MacSpeech Scribe, an advanced personal transcription software package for the Mac OS X products. I use a MacBook Pro for a lot of my work.
My experience with the earlier dictation software versions for Mac OS X have never been as good as my experience with a PC. MacSpeech Scribe convinced me that Nuance rules the Mac world also.
MacSpeech Scribe accepts transcription from electronic devices. MP3 recordings work the best for me, but there are other ways to input (.wav,.AIF,.MP 4, etc.) that work just fine.
MacSpeech Scribe seems to work best when one uses the iPhone as a dictation device, and all iPhones have a utility where one can record a voice memo and sync it with iTunes. MacSpeech Scribe took over from there. It had approximately the same accuracy as Dragon NaturallySpeaking Premium Version 11.5, including its ability to recognize spoken punctuation.
The ability to use the iPhone for dictation is a great advantage. First, one does not have to worry about clarity, especially when it comes to background noise. Second, it's one less thing to carry. For short dictations like e-mails, I use Dragon Dictation on my iPhone. For full-blown articles, I use MacSpeech Scribe.
I don't know of a better product out there then Dragon NaturallySpeaking Premium version 11.5. I do not know of a better transcription product then MacSpeech scribe. I do know that I could easily demonstrate the advantages of using these products in public safety. If I speak highly of the product, it listens.