Nov 28 2011

“Siri, will speech recognition ever take off in Legal?”


Last week I attended the Bighand user conference at the excellent Renaissance Hotel in St Pancras (take note certain legal IT company whose only user event I attended the previous week). Rather than write up a review of the whole day (there’s a good one here if you’re interested) I thought I’d comment on an item that was high on the days agenda.

Speech Recognition.

Now before I get accused of following certain people or the current trend generated by SIRI let me first point out item #1 on this blog post of mine from the 1st January 2010!

But the feeling I got from the conference is that finally the tech, that has been around in Legal ever since I’ve been in this vertical, is finally reaching a point that it is useable. The latest version of the Bighand product (4.2) uses the new Nuance 11 engine and from the demo shown on the day is impressive (demo online too). The workflow with transcribe and proofing seems ideal and the tools given to the secretary to control the dictation playback with the resulting document for amendment is well thought out. I seem to recall in a previous Bighand session that this correcting by the secretary would help with the teaching of the speech recognition software for that author (I could be wrong on this one so check with Bighand first!)

With a bit of work with the API that Bighand now provide you could create a great Fee Earner interface from the DMS (document management system) that would ensure the document being created is started on the correct template, filed in the right place and transcribed ready for a secretary to finish the document.

There were some good case studies from law firms who had started to use speech recognition. Stating that the transition wasn’t difficult for existing Bighand users, but the lawyer had to want to embrace the technology (due to the initial time taken to teach the system and perhaps having to adapt dictation style for better results). Also feedback was not that this led to reduction in secretaries (those lawyer-secretary ratios were high enough already!) but to enabling the secretaries to do other work for the lawyer.

The key point that stood out from the day though were some comments generally on the technology. For a while I’ve thought that maybe dictation was a dying technology, after all the “younger” lawyer is used to typing his/her own documents right? Well this generation maybe, but the next generation is growing up with the likes of SIRI. Maybe this generation is a blip before lawyers throw themselves back at dictation and with the advances in the technology maybe speech recognition is now a viable solution to both improve efficiencies and to make those straining lawyer-secretary ratios work!


Jun 7 2011

Another breakfast with BigHand


Last week I attended one of the Bighand regional roadshows, held again at Gordons LLP in Leeds (I so wish other Legal IT vendors would do these local visits, why should the customer travel down to London all the time?).

I came away thinking there wasn’t much new in the latest releases. But then I guess that’s what a good piece of software should do, tweak a little but don’t add so many features that you ruin the core thing your software does well.

But there were a few new “tweaks”, some of which I’ve highlighted below. The information here is from my notes rather than any official press release, so if you like the look of a feature you may want to check with Bighand rather than trust my note taking! (For info, the notes were taken using the excellent OneNote app on the new Windows Phone 7, which auto syncs with the OneNote webapp via Microsoft’s SkyDrive. I’m drafted this blog post using said webapp on my PC and am finishing off using OneNote in Office 2010 – all integrated beautifully!).

New in v4 of Bighand.


Profiling is available in v4. Similar to how a DMS (Document Management System) profiles documents you can do with your dictations. This then builds up title (saving the lawyer typing up a random title). This profile information can be populated from other systems e.g. Client matter information from the practice management systems. This profile information is then available in the BlackBerry client also.

Practice builder

An additional product that allows for a “drill down” interface as the starting point to get into your dictation. You select your client, “drill down” to the relevant matter and then start the dictation. It’s all about trying to make it easier to build a title without need to type it all in.

At the moment an xml file is used to populate the profile and “drrill down” information. But they say they are looking at a SQL connector in later releases.


These have been expanded to allow multi-step workflows. For example, before a dictation is routed through to the secretary it could go through speech recognition and then a proofing step. It makes more use of the “attach file” functionality, allowing a much more complex (and useful?) workflow to be built. There are plans to look at hooking in 3rd party steps, although an SDK for the workflow is not yet on roadmap so it’s early days in the thinking.


A small but pretty neat addition is allowing the adding of bookmarks at points in the dictation. So a fee earner could, for example, add a bookmark to ask their secretary to check a name as they’re dictating. Or a secretary could add things to clarify with the fee earner. This bookmark information is displayed in the application as each bookmark is passed when playing the dictation. It could be very useful in conjunction with the new multi step workflows.

Send anywhere

This function allows you to send a dictation to any secretary or department without the need to add a specific workflow to do it. It can still be configured to how you want.

There is also the ability to easily set dictation to confidential without need to set up specific options for this. This goes straight to the assistants folder. If a co-ordinator can see that particular workflow they would only see an item titled “confidential dictation”, but they would not be able to open it.


Improvements in speech recognition

The average processing time that Bighand see for 10min dictation on standard specification server is about 6mins (although there was feedback from the room that a firm using speech recognition was seeing processing that was a lot quicker – and they were just running on high power desktop. They were seeing a 30min dictation being processed in 2mins).

Other information Bighand gave was:

  • They fed back that they now have about 1000 customers using speech recognition now.
  • You can spread speech recognition across multiple servers for improved performance, but it does require its own server.
  • Typically it takes about 20-30mins of dictations to get it to learn your voice. There is an addin to Word where corrections are applied, which are then fed back to the system.
  • Works with a number of non English languages. Those mentioned were: German, Dutch, French, Spanish. There is a Legal dictionary available in some.
  • It was mentioned that it requires good training on how to dictate. Requires consistency in volume, consistency in how you dictate etc otherwise get worse recognition rates.

Pre-learning tool

They showed a new addition to speech recognition, the pre-learning tool. This can look at last 7 days of your dictations. You can then play these and paste the doc you produced from the dictation into this tool. This then trains the speech recognition from previous work to your voice. Benefits here are that it would allow a secretary to feed the pre-learning tool on behalf of the fee earner.


Main news here was that an Android client is now available. And in the BlackBerry application there are extra functions in email. Allows you to jump from an email straight to dictate, then when you send the dictation it includes the email from where you triggered the dictation.

iManage integration

There is now a bighand dictation option available on right click of folder within Desksite/Filesite. This then populates the profile in bighand and knows where to create document. It attaches an nrl link to the dictation.  Basically uses the SDK that is available to use. Extra licence through for the SDK.

There was also mention of an addition of client side speech recognition for v4.1, I didn’t take too many notes on this as I see the server side solution as a much better option. Although for small firms the cost benefit of a server-less solution may be better.



Jun 25 2010

Breakfast with BigHand


On Thursday myself and a couple of colleagues attended a breakfast briefing from BigHand at Gordons law firm in Leeds, accompanied by plenty of bacon butties from the Roast! It was one of a number of briefings that they are doing throughout the UK on the back of their recent acquisition of nFlow.

As well realising that it’s not just Herbies that have hot meeting rooms, there was information on the nFlow acquisition. But for the most part we were shown demos of some of the new features being planned for future versions of the BigHand software (I think most were for v3.4). Below are some of the key functions that stood out for me (I was making notes on my touch screen Windows Phone whilst trying to keep up with the demos, so if you’re interested in a specific feature I’d double check my understanding with BigHand!)

  • MS Office Integration. This allows a document to be attached to the dictation and passed through the workflow. Also there is the the ability to create and manage profile information to go with the dictation. These combined allow the Fee Earner to provide information to the secretary on the dictation that the system can then use to, for example, launch a template and fill in details such as document name straight from the dictation in the queue.
  • Then combine this with the SDK and you could enable integration into the DMS (Document Management System) to transmit the document in the dictation workflow, yet maintain the security and version control of the DMS.
  • Escalation function. The ability to set global rules in the system to escalate work. So for example a folder could be given a rule that after a certain time all outstanding dictations are moved to another folder (e.g. out of hours team or the team in the firms Asian offices for example).
  • Reporting has been bolstered by the addition of an analytics module. This stores more information than before in the database (no longer limited to last 30 days) and has improved reporting (no longer using Excel). The module allows you to drill down on results obtaining more detail.
  • In the mobility clients for BlackBerry and iPhone you now have the ability to attach documents and photos to dictations. Taking the devices further into general workflow than just pure dictation (e.g. you could dictate meeting notes that referred to notes on a whiteboard that you’d photographed on the phones camera)
  • Finally the speech recognition module. This was brought in with v3.3 but it’s worth a mention again as it still impresses me. It’s way beyond the old desktop versions, but the addition in the newer version is the ability to chose to either send the transcribed document back to yourself for proofing or send it onto your default workflow for proofing. So the “training” can be done by a secretary checking and correcting your document.

The briefings are still being run, details can be found on their site.


Sep 28 2009

Dictation – Dying out or recession buster?


Some recent experiences got me thinking about dictation and whether it was either a dying art or whether it was a crucial technology to help firms beat costs? First the list the things that gave thought to the fact that dictation is dying out:

  • Comments from lawyers on the differences in the generations. This was actually a discussion on the quality of the dictations, in that the younger lawyers were “embarrassed about their own voices” and dictated more quietly. This made me think that this behaviour may be leading to the next generation of lawyers actually being happier typing up the draft themselves rather than dictating it.
  • Also there does seem to be a number of lawyers now that don’t dictate at all. Again is it easier for them to type up drafts themselves?

But then with digital dictation, especially when the system allows for workflow, I thought there would be a number of added benefits:

  • Devices can be used more effectively for just relaying instructions in addition to dictating documents. Using tools like BlackBerry clients or standard phone lines to route these instructions quickly to secretaries makes this much easier. 
  • Workflow systems generally can make more efficient use of secretaries and document production units, this can be used to smooth peaks and troughs across secretaries (in some cases I guess it can also facilitate use of external companies, as I presume Eversheds recent deal will utilise such technology).

So will dictation die out or take off?

Well I think analogue dictation will die, the days of the tape surely are numbered! But I don’t think digital dictation will die out. Sure there may be some lawyers who can type up short documents as fast as they could dictate them, but for others and for all but the short documents dictation will be faster.

Also with the added benefits workflow brings it will still be beneficial. Given that, in this year especially, the lawyer-to-secretary staffing ratio has increased, the ability to spread the load using workflow will be a big benefit once the growth returns.

The panacea and key technology to secure the future of dictation is of course speech recognition, as it has been ever since I moved into Legal IT. But this is still a long way off in my opinion!

This article over on “The Greatest American Lawyer” that I found after writing this blog post has a good bullet point list of advantages of digital dictation and is worth a read.