By Mike Kelly, Director, Sound Business Systems and based on a recent paper by Derek Austin, Director of Dragon Desktop Solutions, Asia Pacific with Nuance Communications. Thanks also to Mark Robinson SC and Don Thomas for their contribution.
In the opening scene of the 2014 sci-fi movie “Her” you see people dictating naturally to computers with high accuracy and without using headsets. The speech recognition capability shown in that movie is close to what is possible today using the “far field” microphones built into the latest laptops and the latest speech recognition technology.
Speech technology is the most natural way for humans to interact with technology and it is the way we’ve been brought up to think that in the future computing will be done using your voice.
Unfortunately, our computing platforms to date were designed for use with keyboards, mice and, most recently, fingers. Typing, mousing or gesturing imposes interaction methods that do not work well with voice. So voice systems to date have been restricted by the platforms that host them.
We are just starting to design systems where voice is intended to be the main way of interacting
As an example, Jibo is billed as the World’s First Family Robot. It was the most successful technology campaign to date on Indiegogo, the crowdsourcing site.
You could have pre-ordered your JIBO on Indiegogo for $499 with delivery in 2015 but they’ve stopped taking orders. The company asked for $100,000 to get up and going in September and raised nearly $2.3 million!
There is certainly a market for this device and it will be an open platform so that developers can make apps for it. While Jibo may not succeed, it’s a sign of things to come.
Looking forward Speech technology can also help us in the world of work and commerce. The focus moves to the services we will access from our phones, in the car, at home, and in the office. In the short-term, there will be a lot effort to speech enable services that we use every day to make them more efficient. You can already do this today.
Nuance’s Nina is an example of how this technology works.
There are two types of speech recognition software today…
Speaker-Independent Speech Recognition
Speech recognition’s massive advances have made the technology ubiquitous. There is the speech recognition that you use in your smartphone, your game or maybe your TV or alarm system.
This is speaker-independent technology: It works the same for everyone. This is what your smart phone personal assistant uses, with the recognition done in the cloud rather than on your device.
These virtual assistants can help you navigate your life, perhaps do a web search and send a text message. It is what we call Command and Control SR technology.
This sort of capability has always been available on the desktop using Dragon but there is a trend to integrate
this function across all devices and across social media. Dragon Assistant is an example of how this is being done.
Speaker-Dependent Speech Recognition
Then there is speaker-dependent technology like Dragon which is transcription focused and uses speech as a fast and natural way of interacting with a computer and producing text.
Dragon NaturallySpeaking version 13 software, which was released in August 2014, is the most accurate speech recognition technology available today. Dragon is designed for people who create documents, i.e., those who write, whether that be long documents or a myriad of short emails.
Using speech is without doubt the fastest way to write – you can enter accurate text at least twice as quickly as you can by typing, saving you time and money.
It is this second area that has relevance for the legal profession, efficient text transcription being the bread and butter of all lawyers.
Dragon speeds the process up by first making a profile for you as a user and then learning not only how you speak but also what you say and in what context you say certain words or phrases. Over time, if you correct it when it makes a mistake, Dragon will learn and become even more accurate.
High Accuracy
Speech Recognition software is now highly accurate, with accuracy levels of 98-99% quite commonplace. This is clearly not ‘hit or miss’ software that requires great time or effort to master: it is both easy to use and accurate to an almost uncanny extent.
At such levels, it will only be a quantum jump in natural language or context understanding that will allow the accuracy of SR software to further improve. Lawyers can, however, expect innovation around better support for document workflows, the handling of surnames and place names more efficiently and the blurring of the lines between desktop and mobile.
WAYS TO DICTATE
Especially for lawyers who traditionally have dictated their documents relying on back end secretarial support, speech recognition is now so accurate that it allows lawyers to take full ownership of their work and handle the entire process independently.
Particularly for the more recent law graduate who is familiar with MS Word, they find the entire editing and formatting process a very simple procedure. Many lawyers also prefer the immediacy of front end speech recognition.
You can use Dragon “live” by dictating directly into your word processing software or you can use Dragon to process an audio file that you have recorded using a digital recorder or a smart phone application. Correction of the transcribed text may be done by you or by an assistant. Corrections can be fed back into your Dragon profile to improve performance.
If you have an Ultrabook laptop or a Mac, you can use the built-in microphone if you wish although a noise-cancelling headset will still give you the best results. You can switch between them too if you want to use both.
Note that Dragon can’t handle transcriptions of multiple speakers. The software can’t handle meetings, interviews, and the like.
There are a myriad of audio input options to choose from including hands free wireless headsets and units that integrate your desk phone, mobile phone and computer so that you can use one headset for all devices.
There are also hardware and application choices you can make if you want to go the recorded dictation route. In particular, Olympus and Philips make hardware that integrates with Dragon Professional to automate transferring audio from the recorder to your PC for transcription. Dragon also offers an Auto Transcribe capability where audio files can be dropped into a folder for Dragon to process.
Choosing a Voice Recognition Product
For the vast number of users I’m recommending Dragon, but your alternatives are the built in speech recognition features in the operating system you use. To date, these have offered inferior performance particularly for New Zealanders and New Zealand lawyers in particular.
Note that there are different Dragon products for each operating system. Dragon NaturallySpeaking is for Windows. Dragon Dictate for Mac is, obviously, for the Mac OS.
Dragon Dictation on iOS or Android is altogether different and offers a cloud-based speaker-independent service. You don’t already “have Dragon” if you’re talking about your phone!
Training and Customisation
For use within a legal practice Nuance recommends that you work with one of their specialist partners for help in configuring Dragon to suit your workflow and for the all important user training and after sales technical/user support. While Dragon is accurate out of the box it still needs training to achieve best results.
Most Nuance certified partners can provide many of these services via the web these days so they don’t need to come into your office and hours can be flexible. Partners can also advise regarding hardware accessories like wireless headsets and digital recorders.
In New Zealand, Sound Business Systems is Nuance’s most experienced business partner and we have extensive legal industry experience plus an extensive user base you can speak to if references are important.
INSTALLATION and TRAINING
Dragon installs like any other software onto your Windows or Mac system but I recommend that you seek help from a Nuance approved Dragon business partner especially as you will be needing them for user training and technical support.
The software can be installed to work over a network environment allowing you to be able to work from home and from laptops and remote into your Windows Server environment and dictate straight into Outlook/Word.
The first time you use Dragon, it will set up a profile for you. This includes information about what you sound like (the acoustic language model) and what you write (the language model).
Initially, these are based on samples of what New Zealanders generally sound like and what New Zealanders or New Zealand lawyers in particular generally write. You can then train Dragon with your voice and on your documents to learn more about your personal requirements.
Once installed and trained, this software will allow you to dictate all your correspondence via Email directly into Outlook and the same with updating your document management system and any application that you would normally have full key board functionality control.
Dragon integrates seamlessly with the entire MS Office suite and with any Windows compliant software package and allows easy use with Templates, Precedents and documents containing Form Fields.
When Dragon makes a mistake, you can correct it and, over a few weeks, it can become very accurate.
When you create the profile, Dragon will check your audio input, give you the opportunity to read to it for a few minutes (recommended) and then take you through an introductory tutorial on basic commands. Then you are up and running, ready to start using Dragon.
Although Dragon provides accurate recognition out of the box we cannot over-emphasise the importance of correct training from the outset. Not only will you need guidance in order to understand and unlock the efficiencies that Dragon provides, but it is especially important that you also learn how to correct mistakes properly so that your voice profile improves rather than degrades!
You are making a reasonably significant investment when you decide to purchase Dragon in order to improve your efficiency and productivity, so why put your investment at risk by scrimping on training and support?
Depending on requirements, your Dragon partner will start with a system installation on each lawyer’s PC and configure Dragon to the individual work and dictation styles, with auto insertions and a thorough scan of 2000-4000 completed documents. This will ensure that Dragon is personalised to your particular work type and deliver maximum accuracy.
Your Dragon partner may also write custom voice command scripts for your sign off clauses, plus assist with scripts for your practice management system and any other routine repetitive tasks.
Getting up to Speed
The first thing to practice is turning the microphone on and off. You can use the mouse for this but a keyboard shortcut is recommended (Enter key on a full sized keyboard by default).
Our usual recommendation is that you focus on making Dragon work for you by becoming adept at dictating and correcting text. Dragon can be used to format text and pretty much run your whole computer. We suggest you leave that alone to begin with unless you need it for accessibility reasons.
Start with a reality check if you think you can type faster than you speak: How fast do you really type? This may motivate you to get started with speech recognition and keep going.
Tips From Lawyers Using VR
Don Thomas is Principal of THOMAS & CO a leading mid size law firm based in Auckland.
Thomas & Co regularly review IT systems and software to ensure the firm is providing the most efficient service delivery. To implement Dragon onto the firm’s system, they used Dragon partner Sound Business Systems for support, as well as one-to-one training and technical support.
“Our lawyers took to Dragon quickly because the software is so intuitive and its implementation had a major impact on the firm. Rather than speaking into hand held digital portables, our lawyers now speak to their computers and Dragon quickly transcribes their words with a 99% level of accuracy.
“It has also made editing easier, as lawyers could alter the words as they appeared on the documents. Free to focus on the content, Dragon has helped my firm to speed up the document creation process and to maximise
productivity.
“Thanks to Dragon we assumed complete ownership of our own documents from beginning to end. There is no more queuing and waiting for documents to be typed up and returned. Each author can now produce and edit their own material without typing, achieving an immediate and a far quicker and easier turnaround than ever before.”
“Dragon has also allowed us to redirect our resources into other fields. Now that we are able to produce our own documents on the spot, this has freed up our typists and secretaries. With the need for typing now defunct, we have been able to redirect them into more productive areas as true personal assistants and authors in their own right.
Now an embedded and essential tool at Thomas & Co, I have recommend Dragon to other colleagues in the legal fraternity. It provides excellent value for money and has helped us become more efficient, cut costs and save time. From a four author six assistants structure we have changed to six author two assistants with more turnover.”
Tips From a Senior Barrister
Mark Robinson SC is a long standing Sydney based senior counsel and Dragon user based in Sydney specialising in administrative law and general law.
“I would train the software and be patient enough to correct the dictation each time it gets it wrong. Always take the trouble to train names. As to names you will never say a second time, just stop dictating, and manually type “Mrs Lucy Kodszwoski” then start dictating again. She will probably be called the client, the plaintiff or the applicant thereafter – easy.
“I would also back up everything and keep on back-up off site at all times.
“Soon enough, correcting text and creating saved words (and backing up) will become second nature to you and your overall performance will certainly improve. Even if it does not, Dragon is better and faster than typing.
“Very soon, within the first week, you will establish the dictation flow that works best for you. Stick to it. My dictation style is probably a little slower than most, but I like to see the words appear as I go and correct each sentence as I say it (if it needs any correction at all). I save my voice profile often in this process. That way, I generally dictate and review at the same time.
“When I am finished (usually a summons, originating process or submissions) I then print and sign it. This is one of the greatest benefits of the software, I only touch a matter once. I do not have to review it for content or accuracy. I can forget about it from then and there.
“If I had more time, I would create more macros although I know it is a relatively simple procedure. The only time this takes is in thinking up names for them that you will remember. I write them down in an exercise book.
“The most helpful thing to be alert to is the Dragon sidebar command feature called “What Can I Say?” If you ever get stuck, a context sensitive dialogue box appears on the right hand side of the screen to remind you of the available voice commands particular to the program you are currently working in.
“If you can’t remember, just say “What Can I Say?” It is very useful. If you are in MS Word – it sets out those commands. If you are in Microsoft Outlook, it displays those commands.
The following are voice recognition tips from Derek Austin of Nuance.
- Dictation is a skill that can be learned but which sadly is no longer taught to young lawyers. Barristers are rarely deficient in the area of speaking ability but dictation may present some challenges. If you are not adept at dictation, we recommend daily practice as a way of coming up to speed.
Here are some tips on dictation: Dictation – A Complete Guide – How to Dictate Effectively
- Remember that you do not have to use Dragon or the keyboard and mouse. You can use both. So do what makes sense for your style. Many lawyers produce a first draft of a document using Dragon and then format it using the keyboard.
- Training and customization is recommended. That’s where our approved business partner provided value.
- Dragon Professional software lets you make simple templates for easy navigation between information fields using the “next field” command. You can also store your profile in one place and have it synchronised between your office computer and your laptop.
- Voice commands can automate entry of often-used text.
- Edit your Dragon’s vocabulary if there is something in it you don’t like add words if they are missing, and delete ones that are misrecognized
- Check your audio settings when your mood changes. Check them again when your mood changes back. Also check audio settings when you are in a new location for maximum accuracy.
- Bookmark this page in your browser. From here you can download shortcut summaries, instruction manuals, and get information about how to call Nuance’s support team.
The following links may also be of interest:
Information about Dragon software
The post Speech Recognition…a Technology That Makes a Difference appeared first on LawFuel New Zealand.