My experience of trying voice recognition software again after a failed experiment some years ago, has messages for those who have not caught up with developments in litigation support software.
I have come back to voice recognition software after many years of assuming that it was an unwieldy and inaccurate method of transferring words from head to screen. I am immediately hooked and regret all those years spent crouched over a keyboard. Or do I? Is it possible that I have come back to it just at the point where it has reached a level of accuracy which is adequate for my needs, and just when those needs are greatest?
The e-disclosure context here is all those potential users of litigation support applications who dip their toes into the water once and retreat vowing never to try again. Some inadequacy, ranging from an outright system failure which lost their case through to a minor annoyance which became too tiresome to tolerate put them off, often with cries of “I told you so” ringing in their ears.
Despite my long-standing interest in technology, I am not generally an early adopter — not any more. Does anyone else remember Windows 2.0? How about the upgrade from Windows 3.1 to Windows 95? Anyone who went through that and is still a computer-user deserves marks for persistence – a real triumph of hope over experience. The same is true of many early adopters of electronic disclosure.
I tried voice recognition when it first became commercially available. It was expensive and required an immense amount of training – that is, training the system to recognise the user’s voice – and hours of this training seemed to produce little improvement. I was about to give up anyway when a disk drive failure destroyed the product of the training. I was not too sorry.
In those days, I was a software developer and most of my time was spent coding, testing and supporting software, none of which required much typing. There was not a great deal of e-mail traffic then, either, and although I wrote the occasional article, my untrained typing fingers could keep up with my thought processes. It is rather different now – most of what I do is writing, and I produce between 3,500 and 5,000 words in a typical week. The output so far this week is over 10,000 words without counting e-mails and amendments to existing documents.
In anticipation of this, I bought a copy of Dragon NaturallySpeaking 10.0, enticed by the promise on the package that little training was required. It was an impulse purchase – I saw it on the shelf and took it to the till without my usual research into alternative products, comparative pricing and all those other brakes which we invent to defer a decision. I had to read one passage to it and chose J F Kennedy’s inaugural speech which was no great hardship. Then I started my first document. It just worked and I was away.
I am a slow writer, rolling sentences around in my head before typing them, and revising paragraphs as often as it takes to be happy with the flow and rhythm. The main benefit to me of voice recognition is not the advertised ability to type as fast as I think but to try out ideas without having to type them. There is a lesson here, perhaps, for software suppliers: you sell applications to people by finding out what they want to do and enabling that, rather than by telling them what they should do.
I wonder how you manage with voice recognition software if you are of the generation which drops all its consonants and elides its vowels. Dragon claims to deal with regional accents but I cannot imagine how it copes with the un-accented slurring of Estuary speech. Not everything came out as I intended – I am glad I spotted that my reference to Professor Hazel Genn’s “Hamlyn lecture” was typed as “rambling lecture”, which I am sure it was not; my mention of “Greece making runs” would surprise cricketing enthusiasts, as it did me – it was the Victorian cricketer W G Grace to whom I had referred; I rather liked the idea of “kids managing judges” although the intended “case managing judges” better suited the context. On the whole, the failure rate was low and it is easy to train the system to recognise words you are likely to repeat — Digicel comes up quite often in my writing, for example, so I have added it to the vocabulary.
This application has transformed my working life, and not just in respect of the long and careful documents. E-mails, which generally do not need the same editing attention, are much easier to rattle off. I can talk to my precious CaseMap database, full of cuttings, bank statements and accounting records, instead of typing into it. I use Microsoft’s OneNote as a scratchpad for dumping unstructured thoughts to which I may return one day, and its value is substantially increased by being able simply to click on a new page and start talking.
So, to revert to my starting point, have I wasted years typing when I could simply have been talking at my computer, or have I been well out of it whilst the technology was developing? It does not matter really. I tried the application years ago and did not persist because the technology was not adequate and I did not need it enough to keep at it or to try again. I have tried again and it now perfectly matches my needs.
I do not rabbit on about my own working practices without a purpose. Many lawyers who have considered the use of litigation support technology in the past will have reached the same conclusion as I once reached about voice recognition – they could not see a need for it which justified either the cost or the investment of time needed to make it work for them, it was not easy to use, and it lacked functionality. The relationship between need, cost and time has changed, as has the quality of the software applications. The need follows from the volumes of documents which exist at every corporate client, from the pressure to keep costs down, and from the outcome of last year’s e-disclosure cases, Digicel, Abela and Hedrich. The cost of services and software has come down, partly because the early adoption phase has ended and partly through competition. Time is the most expensive contributor to the cost of litigation, and anything which reduces the hours has to be considered if firms are to compete for business. Lastly, the power of the current generation of litigation software applications, and in particular those designed for early case assessment, has to be seen to be believed.
A few years ago, the acquisition of any litigation support system generally involved significant capital expense and staffing implications. That is no longer true. Although you can buy applications for in-house use, you can also rent them, whether for a one-off case, for all your cases or as a step on the way to an in-house purchasing decision. You will not quite pull one off-the-shelf on impulse, as I did with my voice recognition package, but the decision to run a case electronically involves far fewer implications than it used to.
The array of logos down the side of this article is the equivalent of the shelf from which I picked my copy of Dragon NaturallySpeaking. Clicking on them equates to picking up the box and reading about the functions and benefits. You will not generally find a price tag, but you will find contact details. If you last looked into the use of litigation support software say a year ago, you will find changes in the range, the functionality and, not least, the price. I doubt that you will make an impulse purchase, but you will find it much easier to relate what the suppliers are offering to what you need to do.