How to Use Speech-to-Text Dictation for a Quantum Leap in Efficiency

The feeling you'll get when you use speech-to-text dictation.

The feeling you'll get when you use speech-to-text dictation.

I finally removed the small impediment that was preventing me from fully implementing speech-to-text dictation, and I have realized a quantum leap in efficiency as a result. You owe it to yourself and your clients to take a few minutes to read this article, in order to see the amazing things you can accomplish with this technology.

I'm not talking about dictating files that are then given to your secretary for transcription (although you can do that as well). I'm talking about software that allows you to create documents with your voice, instead of the keyboard. If you are resistance to voice-to-text dictation software, it may come from the fact that your experience was similar to my own.

The only viable dictation software I have experienced is Dragon NaturallySpeaking, by Nuance. (I'll just refer to it as Dragon.) There have been other programs, and Windows has included speech recognition/dictation since Vista, but from my experience Dragon currently owns this niche. I tried Dragon once long ago (it actually came bundled with WordPerfect at one time) but never really got into it. It required an investment of time to “train” it to your voice, and even then it was only about 90% accurate, and the time spent editing out all the mistakes it created ended up being comparable to what it would have taken to type the document in the first place. In my case, I’ve gone through that process of trying to make it work a couple of more times. I was sent a review copy at some point, and even paid to upgrade it to keep it current, but never got into the habit of using it. However, I finally forced myself to take the time to implement it and to develop my technique, and my efficiency has soared. I am a very fast typist, but I cannot type anywhere near as fast as I talk (no one can, that’s why we pay the big bucks to court reporters, and even they must constantly tell people to slow down). Dragon claims that you can input text with your voice about five times faster than with a keyboard.

With legal briefs to write and multiple law blogs to feed, I need to be able to input text as fast as Einsteinian Physics will allow. I just talk into my computer, even gesturing as though I am explaining something to someone, the only difference being that I end sentences with punctuation and say “new paragraph” when appropriate. I usually don’t look at the computer monitor during the process. I'm too tempted to correct a mistake if I see it, and I found that the time it took to edit by voice (“select ‘Brad Berry’, spell it, B R A D B U R Y, go back”) was the time-consuming part of the process. If I make a mistake, I just say “fix this” and keep going. (I used to just say “crap”, but I didn’t want that to mistakenly end up in a document.) I just get it all into the document and then go back and edit.

You like I’ve modified my technique slightly as I’ve gotten more comfortable with Dragon. If I pause to collect my thoughts, I take a look at the screen and scan what I’ve just dictated. If I see a mistake that just involves replacing a word, I usually fix that with Dragon because it is so fast; faster than using the keyboard in most instances. For example, if I dictated the word “collage” and Dragon heard it as “college”, I just say “correct college”, and Dragon instantly highlights the word and offers a list of all the similar sounding words. Usually, the word I intended is in the number one position. It's as though Dragon thought I might be saying that word instead of what it used. All I have to do is say, “choose 1” and it’s done. It corrects the word and returns me to where I was in the document.

Writing an article with the keyboard now seems so last millennium. It’s like the scene in the Star Trek movie, where the crew travels to the past (our present), and Scotty sits down at a computer and starts talking. When that doesn’t work, he picks up the mouse and talks into that. Finally, when someone points to the keyboard, Scotty says, “how quaint.”

I use Dragon on my desktop computer at the office and at home, as well as on my laptop. Multiple installations are permitted under the software license:

Dragon is licensed on a “per individual” basis. You are permitted to install the software on more than one computer (such as on a desktop and a laptop computer, or on a work and a home computer), but you cannot use the software concurrently on more than one computer. You are permitted to create multiple voice profiles, so long as each voice profile is for you.

I've been using Dragon NaturallySpeaking to a limited extent since version 10, but when I fully incorporated speech-to-text into my practice, Dragon 12 was the current version, which was released in October 2012. Nuance thereafter offered a free upgrade to version 12.5, and either because I was getting more comfortable with the software or the technology got better, that is when it really seemed to hit its stride. The misuse of words (using "threw" when I meant "through") almost never happens. The accuracy is phenomenal. Even though I've been using it for some time now, I am still amazed by the program. Sometimes, even when I mumble, I'll look up at the screen and see that Dragon got it right. As an example of how good the program has gotten, when I just dictated the word "Dragon", it knew to use a capital D. Somehow, contextually, it knew I was talking about the program, and not a dragon.

Nuance released version 13 in August 2014, thereby rendering version 12 "obsolete" and turning it into a great bargain for those wanting to try dictation. Every time they come out with a new version, they tout the improved speech recognition, but version 12 is so good that it’s hard to imagine that version 13 could be appreciably better. If you want to try dictation, my suggestion would be to buy version 12 (it will automatically upgrade to 12.5) at the bargain price for now. You can always upgrade later. In fact, you can get version 12 with a digital recorder for about what you’d pay for version 13 without the recorder. BUY THE PREMIUM EDITION, not the "home" edition, because the latter lacks some nice features. As I write this, Dragon NaturallySpeaking 12 Premium Edition is about $75, and Dragon NaturallySpeaking Premium 13 is about $175.

By the way, I'll be discussing Dragon's ability to transcribe audio files in a minute, and Nuance usually offers a bundled "mobile" version that includes a digital recorder for that purpose. If you are considering going that route, be sure to independently research the offered recorder, first to make sure that it is a quality recorder with good reviews, and second just to make sure you couldn't buy the same recorder separately for far less than the increased bundled price.

For the record, I'm using yet another version referred to as Dragon Professional Individual. Nuance has dropped the "NaturallySpeaking" moniker, and just refers to this latest version as Dragon. At $300, this is obviously a more expensive version, but it has a feature that I think makes the extra money well spent. You can create word combinations that then insert whatever text you have designated. For example if I'm dictating an email, when I say "standard signature block", Dragon inserts all of my contact information and the disclaimer. I've included a video, demonstrating the use of this feature, following this article.

I've just started using this latest version, so I have not yet taken full advantage of this feature, but I see that it is going to be a huge time-saver. For example, when I create a demurrer, I pretty much use the same three paragraphs every time that set forth the legal standards for a demurrer. I can now set it up so that when I say "demurrer law" Nuance will automatically insert those three paragraphs. I could accomplish the same thing with any text expander program, but this is super efficient.

Be sure to click around on Amazon to see what is available, because there are a number of permutations (including MAC versions). For example, at the time I am writing this, you can buy Dragon Professional Individual for $300, but there is a special Dragon Professional Individual BLUETOOTH edition that comes with a Bluetooth headset for the same price. Alternatively, if you take my suggestion and buy version 12 for $75, and decide you like dictating your documents, you can then Upgrade from Premium 12 to Dragon Professional Individual for $150, resulting in a savings of about $75 over the $300 price. From what I have seen, Nuance upgrades go back two versions, meaning that if you buy version 12, you'll still be able to upgrade it at the lower upgrade price through version 14. Also, once you register a Nuance product and are on the mailing list, they offer some real deals.

There is a Dragon NaturallySpeaking Legal edition that is about $700, but so far as I can tell the only added feature is a built in lexicon for lawyers. For my purposes, I would have no reason to spend that extra money. You can teach Dragon any word, and if it is a unique word that is unlikely to arise in the future, such as a legal term or someone’s name, I use a simple word in substitution, and then do a global search and replace when I go back to edit.

You can do more than dictation with these programs, including operating your computer. For example, you can open your email program, dictate an email to a client and send it without ever touching the keyboard. For me, though, I just use it for the lightning quick entry of text; mostly for articles and legal briefs, and to respond to emails. For court documents, where I will be cutting and pasting research and moving around arguments, I still find that the “quaint” keyboard is the way to go, but even there I use Dragon to speed up the process. To start dictating, you just say “wake up” and the mic turns on and Dragon types whatever you say. So, when I am preparing a brief and come to a point where I am going to insert my brilliant analysis, I just say “wake up”, dictate in all that I want to say, and then say “go to sleep”, at which point I return to typing.

And then there are deposition summaries. For you litigators, using Dragon to prepare deposition summaries is like discovering pen and paper when previously you had always carved your words into stone tablets. I just kick back, often with my feet on my desk, and read the deposition transcript. When I read something that I want to include in the summary, I just say, "23 colon 12 hyphen 17 tab Jones never attended Trump University period new paragraph 23 colon 12 hyphen 17 tab Jones is unaware of any information that would support the negative comments he posted in his Yelp review period", and it is printed as follows (but lined up better than I can achieve in WordPress):

23:2-5 Jones never attended Trump University.

23:12-17 Jones is unaware of any information that would support the negative comments he posted in his Yelp review.

I'm sure the input method sounds a little strange to the uninitiated, but it becomes perfectly natural, and I estimate I can summarize a deposition transcript in about one-fifth the time it used to take me when I wrote them out by hand or typed them.

Dictation also beats the heck out of OCR.  A good OCR program is a wonderful thing when you have only a printed document that you need to get into your word processing program. But unless you have mad OCR skills that I lack, it is always a frustrating process, especially if I am scanning a court document with pleading numbers. I am apparently not alone in this experience, because when I ask a legal assistant to OCR a document, the usual response is, "can I PLEASE just retype it?" With Dragon, if the portion I need is just a page or two, I can just read it into the document in far less time than I could OCR it, with better results.

Here is a good video that shows Dragon in action with some basic text creation. This video is for version 13, but it is just the same for the other versions.

Notice in the video that she is dictating directly into the documents (Word, Outlook) she is composing. That is how it works for most programs (even WordPerfect!), but if you call up a program that does not link to Dragon, it just opens a dialog box. You dictate what you want, click "transfer", and the text you just dictated is pasted into whatever program you are using.

A few words about headsets and microphones.

Some of the versions of Dragon come with a headset. The quality of the headset you use will directly impact the accuracy of the dictation (to a point). As you can imagine, the quality of the headset included with the software is going to be a compromise between price and quality. I've tested a number of USB headsets, including the one that comes with Dragon, and a Bluetooth headset. I really liked the Bluetooth headset because I could pace while dictating (it seems to help my thinking), but it appeared that the accuracy suffered slightly. (This was not the Bluetooth headset that comes with the Bluetooth edition of Dragon, so I can't speak to the accuracy of that headset.) Also, the Bluetooth headset created an annoying extra step. You can't just leave it on because that drains the battery. So, to use it, you have to turn it on and wait a few seconds for it to link with the computer.

Thus, for me, if I'm going to use a headset, I prefer a wired, USB headset. Out of all the USB headsets I have tested, I achieved the best results from the Logitech ClearChat USB Headset H390, which is less than $35.

I mentioned above that I removed the final, small impediment to dictation. I have long used voice-to-text dictation as a supplement to typing, but now it is becoming my primary input method, because I have eliminated the need for a headset. Although headsets presumably give the best accuracy, I don't like wearing a headset all day, so the headset was providing just enough of an impediment that I would keep typing rather than to take a few seconds to put on the headset. Back when I first started using dictation software, the quality of the headset was essential and, in fact, Nuance published a list of "approved" headphones that would work with its software. Even though I have become a complete zealot as regards dictation, I was still operating under the assumption that a quality headset was essential. Then I happened to notice, while I was upgrading to Dragon Professional, that the instructions included details on how to use your laptop's internal mic. "Well," I said to myself, "if I can use the internal mic on my laptop for dictation, why can't I have a microphone in front of my monitor when I'm using a desktop computer?" It turns out the answer to that question is, "you absolutely can." I now have a microphone on a desktop stand, just to the left of my keyboard, and the accuracy appears to be just as good as with a headset, which is to say, very good.

With this configuration, speech to text is seamlessly incorporated into all that I do. Before, I thought in terms of whether I was entering a dictation session. If I was about to prepare a brief, I would put on my headset in order to be ready for any dictation. Now, I've eliminated that thought process. The microphone sits there patiently, waiting for me to say "wake up" and dictate something. I'll use Dragon to dictate a single sentence, something I never would have done if I needed to don the headset. Very often, using Dragon is far better than using the keyboard. For example, I can use dictation for more challenging passages, because I don't have to worry about any tricky spellings.

Watch this. Supercalifragilisticexpialidocious. I knew Dragon would know that word, even though WordPress marks it as misspelled, and offers no suggestions.

At home, I use my podcasting microphone, which is a Audio-Technica ATR2100-USB, which is probably overkill, but I already had it so why not use it. For the office, I ordered an ELEGIANT SF-920 Multimedia Studio Wired Handsfree Condenser Microphone, which is just a cheap $13 condenser mic, but it comes with a stand and looks totally cool on my desk and works great (see photo to confirm the coolness factor). Ironically, I bought this mic thinking it would provide a test of the low end of the spectrum. I already had the Audio-Technica, to test what a quality mic could accomplish with Dragon, so I wanted to see how cheap you could go and still achieve acceptable results. As it turned out, this unassuming little $13 condenser mic is rated very highly. But, while my low-end test was thwarted, the point is moot if you can buy such a quality mic for so little money. Incidentally, as you can see from the photo, this is not a USB mic. Be sure your computer has a convenient analog microphone jack before you buy this mic.

[Bonus: I've been using Skype more and more to meet with my clients by video conference, and I've been told repeatedly that the sound quality is exceptional. Typically when making a Skype call, the mic and camera are one unit, so the mic is some distance away. With a desktop mic, I can move it much closer to my mouth and thereby eliminate the echo.]

Let me be clear though. A headset has one big advantage over a desktop mic, in that it moves with you. Your mouth needs to be a consistent distance from the mic for the best results with Dragon. With a headset the mic is always the exact same distance from my mouth, even if I am moving around, reading from documents, etc. With a desktop mic, I basically have to be facing forward at all times (although a condenser mic is far more forgiving of a little movement than a dynamic mic). If I'm heading into a heavy-duty dictation session, I still don the headset, but having the mic on my desk makes dictation available to me at all times. Like me, I imagine you'll end up using both.

You can dictate even when you are away from your computer.

The efficiency does not end with using speech-to-text dictation at your computer. Dragon doesn't limit you to real-time input with a mic. One of the reasons that you want the "Premium" version of Dragon and not the "Home" version is that Premium will accept dictation files (as does Dragon Professional). Dragon for Windows can recognize .wav, .wma, .dss, ds2 and .mp3 files. Dragon 13 and Dragon Professional also accept m4a files, which is the native file format from iPhone voice dictation. Dragon cautions that these must be in your voice, because the profile is optimized for you. I haven't tested the claim, but it rings true. I dropped one of my podcast files into Dragon, and was amazed by the accuracy of the output, albeit without any punctuation, because obviously I don't articulate how a sentence should end during a podcast. So, even when you are away from your computer, you can dictate a letter or whatever, and dump it into Dragon later. The challenge, though, is how to accomplish that "dumping" process.

Dragon Recorder.

Chances are you already have a portable dictation device right there in your pocket or purse. Any smartphone has a built-in app for voice memos, or you can get the Dragon Recorder app for free. With the right set up, you can just plug your smartphone into your computer with a USB cable, and it shows up on your computer as an external drive. You can then just point Dragon to the file(s) you want to transcribe.

For me, that's too much work, and it has an added complication if I'm using my home computer. There, when I plug in my iPhone, it opens iTunes and syncs my phone. Not a bad thing to happen, but annoying when all I want is to quickly transcribe a file. The Dragon Recorder app avoids all that. When you open Dragon Recorder, all the voice files are there with whatever name you gave them. Assuming your computer and smartphone are on the same wireless network, the Dragon Recorder app sees your computer and provides an IP address for you to enter into your browser. If you enter that IP address, your browser will display all of the files you dictated into your smart phone. Very cool.

Digital Recorder.

I've gone a different direction. Instead of my smartphone, I bought and use a digital recorder for my dictation. The Sony ICD-PX440 Digital Voice Recorder can record more than 1,000 hours (!) of voice memos. But unlike most digital recorders, which connect to your computer with a cable, this one has a built-in USB plug. You use it just like a flash drive; plugging into your computer and then accessing the files directly from the recorder. No uploading necessary. But check this out because it gets even better. The Sony Recorder comes with free software called Sound Organizer that automates the upload process if you so desire. When you plug the recorder into your computer's USB port, it automatically uploads any files. For its part, Dragon will monitor any folder you designate, and automatically transcribe any files if finds there. Put them together, and you have an automated transcription service! Just plug in your Sony Recorder, and you'll get a message from Dragon when the transcription is completed. Amazing. Even though the digital recorder is one more gadget I have to deal with instead of just recording onto my smartphone, it's worth it to me because of the form factor and incredible ease of use when dealing with the files.

Dragon Anywhere.

If even that is too much work, you can take all the file transferring out of the transcription process by using Dragon right on your smartphone. Nuance puts Dragon on your smartphone with an app and service called Dragon Anywhere. It works just like Dragon on your computer, but since your smartphone doesn't have the sort of computing power necessary for speech-to-text dictation, the app uses Nuance's computers via the Internet. (Incidentally, Siri on the iPhone does the same thing; that's why you can't use Siri if you don't have an Internet connection.) Dragon Anywhere will also propagate your auto text and custom words across all your computers and devices. But this convenience comes at a price of $15 per month (less if you buy longer terms). Yes, it would be convenient to be able to dictate and fully edit a document right on my phone, but as I stated earlier, I don't really use Dragon to edit. For that reason, I wouldn't save much if any time over just dictating the document and then editing it on my computer. But if you could make use of the ability to dictate, finalize and send off documents from your phone, this might be the application for you.

I've included a very detailed video after this article, demonstrating how to use Dragon to transcribe files, but note that it does not show the file method I just described.

If I still haven’t completely sold you on the virtues of dictation, invest 99 cents (free if you are a Kindle Unlimited subscriber) and get the e-book How To Write A Book Without Typing It in 6½ Easy Steps. This very short e-book contains a wealth of information about the advantages to dictation, that go beyond the mere efficiency. The author argues, and I agree, that dictation leads to better, more natural writing.

Here is the video I promised, showing the use of the auto text feature contained in Dragon Professional:

6 thoughts on “How to Use Speech-to-Text Dictation for a Quantum Leap in Efficiency

  1. “For example, when I create a demurrer, I pretty much use the same three paragraphs every time that set forth the legal standards for a demurrer. I can now set it up so that when I say ‘demurrer law’ Nuance will automatically insert those three paragraphs.”

    You’d save even more time (and space, if you are worried about your memorandum going over ten or fifteen pages) by omitting those three paragraphs entirely. A number of years ago, I heard then Superior Court Judge, later Court of Appeal Justice, Richard McAdams say that whenever he got to the part of summary-judgment motion stating the standards to apply, he turned the page. Since then, I have never included the standards on a demurrer or summary-judgment motion, and no judge has ever complained about the absence. I have heard of other judges who say that they face more of them in a month than a lawyer will in a lifetime and don’t need to be told what to do.

    I have not missed the post’s point. I may well give Dragon a try.

    • Jake,

      I sometimes teach a litigation course at the local law school, and I drill into my students the exact opposite of what you suggest, based on my own experience.

      You have to dumb down your briefs to the lowest common denominator judge. If you choose to write at some level that assumes the judge possesses a basic knowledge of the law, you will inevitably encounter a judge who is operating below that level.

      Many years ago, I opposed a motion for summary judgment that was easily refuted. The issue involved whether a written agreement had ever existed. My client had a written employment agreement, but made the mistake of keeping his copy in his office. He came in one day to find that he had been fired, and all of his files, including the written agreement, had been taken. Since the defendant employer was now in possession of my client’s only copy of the written agreement, it took the position that it never existed, and filed a motion for summary judgment on that basis. The employer filed two declarations saying the written agreement never existed, but my client refuted the point with a declaration establishing its existence. The law is very clear that this creates a triable issue, and that the motion must be denied on that basis. But come the hearing, the judge stated that since moving party had provided two declarations saying the document never existed, and my client had provided only one making the opposite point, the evidence weighed in favor of the moving party. Like you, I had attended seminars where the judges claimed they don’t need to be told the standards for a motion for summary judgment or demurrer, and since I had not set forth the law stating that a judge cannot weigh evidence on a motion for summary judgment, this judge was clueless on the point. When I explained that the court cannot weigh the evidence on a motion for summary judgment, the judge answered:

      “Of course I can weigh the evidence. The court is the trier of fact. What would be the point of a motion for summary judgment if I had no ability to weigh the evidence?”

      We ultimately prevailed on the case, but I learned that you should never assume the judge knows the standard for a particular motion.

      Another incident also drove home this point. On this occasion, I was attending an arbitration. I represented the plaintiff, who was suing for fraud. The arbitrator was a retired judge, who had previously sat on the bench for more than 30 years, much of that time as the presiding judge. On direct examination, I was taking my client through the misrepresentations that had been made to him by the defendant, when the judge interrupted, sounding like Foghorn Leghorn.

      “Sir, you cannot testify to, and will not abide you attempting to tell me, what anyone said to you outside of this courtroom. That is called hearsay, and it is not permitted!”

      This being a more informal setting, and me thinking that the judge might be up to something I was missing, I leaned over to him and said, “You do know that’s not hearsay, don’t you?”

      “Of course it’s hearsay!”, he bellowed. “He’s testifying to what the defendant said to him. How could that not be hearsay?”

      “Because this is a fraud action,” I said. “He is testifying that the defendant LIED to him, so the statements are not being offered for the truth of the matter asserted. He is testifying that these statements were lies. How could any plaintiff ever establish a fraud claim if he can’t testify to the fraudulent statements?”

      “In 30 years on the bench, no one ever explained it to me like that before. I guess that makes sense. I’ll allow it.”

      So here was a judge who had been on the bench for decades, apparently refusing to let anyone testify to anything said outside of court. No doubt, if he had held a seminar on litigation, he would have said that you don’t need to explain the rules of evidence to him, but he would have been wrong.

      In the specific case of demurrers, I have encountered judges (often to my benefit when I am the one bringing the demurrer) who consider matters outside the four corners of the complaint. It may be that they have considered hundreds of demurrers and genuinely believe they don’t need to be reminded of the applicable law, but there they are allowing a “speaking demurrer”.

      In my never to be humble opinion, you should never omit the legal standards from your motions, regardless of what the judges might claim. You don’t need to provide a treatise, but I think it is essential to always provide the key legal standards for the particular motion. Even if it is the judge’s practice to skip that section because they think they know the law, you’ll have something to point when it becomes clear that they don’t.

  2. The judge’s position is even more astonishing because your client was testifying about the statements of a party to the case. Even if your client was offering the defendant’s statements for the truth of the matter asserted, that would fall under the “party admission” hearsay exception, would it not?

      • Indeed. Early in my career, there were a couple of times when I was absolutely stunned by judges who seemed to think the crux of a particular issue was an underlying area where the law seemed absolutely settled and that the opposing party hadn’t even challenged.

        It didn’t take long for me to learn to build motions by seriously addressing the law behind each building block for a particular argument–no matter how much I thought it was beyond dispute–and to prepare for oral argument the same way. That approach has served me well.

        Your various blogs and online presentation are beyond impressive, by the way. Excellent work, sir.

        • I was once at a legal writing seminar where one of the judges on the panel said that if an attorney uses the passive voice in a motion, he looks for a reason to deny the motion. You just never know what you are up against, but I think it best never to assume the judge will know the law.

          And thank you for the very kind words.

Leave a Reply

Your email address will not be published. Required fields are marked *