John Lawrence (MSFT) : Speech Application SDK

Great whitepaper on the Speech Server Log Analysis tools.

johnlawr — Fri, 07 May 2004 00:29:00 GMT

Written by my esteemed colleague Stephen Potter:

Log Analysis and Tuning with Microsoft Speech Server

Web sites are failing visually impaired users - can the Speech Application SDK help?

johnlawr — Wed, 14 Apr 2004 16:10:00 GMT

The BBC has news of an interesting report: Websites 'failing' disabled users

The key finding was that the group worst affected are those with visual impairments, and the article refers to the recent announcement that the British magazine New Statesman is going to be making its web pages available as speech using new software by Texthelp Systems.

The Microsoft Speech Application SDK ought to make it much easier for web developers to create web pages that support visually impaired users through the use of speech. I would like to think that at some point in the future all ASP.NET web developers considered adding speech controls to their pages to support accessibility.

One of our roadblocks right now, however, is that we don't (yet) have an easy web download of our Speech Add-In for Internet Explorer available from the Microsoft web site, and the redistributable that comes with the SDK is rather large (it includes the speech recognition engine, even through it could use the built in engine in Windows XP ).

If you'd be interested in this as a more widely available installer for the Microsoft Speech Add-In, please leave me a comment and I'll make sure it gets passed on to the appropriate people - we're always looking for input to help us evaluate the importance of features like this.

Eric Rudder mentions the Speech Server launch on his blog

johnlawr — Tue, 13 Apr 2004 16:45:00 GMT

From Eric's blog, The Developer Life

I had a chance to spend much of this past week in San Francisco, meeting with customers, partners, and a few press folks at VS Live/MDC/SpeechTek. It was great to see Speech Server finally launch after so many years of work by our Speech team. I am quite proud of the product they created and I remain incredibly excited by the possibilities offered by natural computing over the long term - it's one of the main reasons I keep working in this industry! I hope some of you get a chance to check out this technology, as it's really a lot of fun to play with!

Thanks, Eric!

Early adopters report benefits with new Microsoft Speech Server

johnlawr — Wed, 31 Mar 2004 21:23:00 GMT

Story in Computerworld about Microsoft Speech Server's early adopters.

My favorite quote:

“By using its familiar Visual Studio tools, Pritchett said, SAICS was able to develop and deploy its application using Speech Server for at least half the cost and in a quarter of the time.“

This is in comparison with developing the application for a competitive speech platform, with other tools. Nice to know we're making people more productive.

Clarification: Speech Server 2004 and Speech Application SDK do NOT require Whidbey/Visual Studio 2005

johnlawr — Mon, 29 Mar 2004 19:47:00 GMT

One of the most frequent questions I was asked at the booth at VSLive/SpeechTek was whether the Speech Application SDK required Whidbey/VisualStudio 2005. I think, this was largely because the other demos during Bill Gates' speech were of features new in Whidbey.

Understandably, people are concerned about developing for a new server platforming using tools that are still at a tech preview stage. However, just to clarify completely:

The Microsoft Speech Application SDK for Speech Server 2004 will run on Visual Studio 2003

You do not need VisualStudio 2005 (Whidbey) for the Speech Application SDK

In other words you can use it right now! You can download Beta 4 here, and we're expecting that the final version will be up in a few weeks.

More VSLive official coverage

johnlawr — Sun, 28 Mar 2004 06:02:00 GMT

Kai-Fu Lee's official keynote: http://www.ftponline.com/reports/vslivesf/2004/lee/

Picture gallery at: http://www.ftponline.com/reports/vslivesf/2004/photos/#

In particular take a look at Bill visiting the speech stand

http://www.ftponline.com/reports/vslivesf/2004/photos/images/08Gates-Hall-1.jpg

See the balding guy in the white shirt at the left? That one's me...

More pictures from the Speech Server launch

johnlawr — Thu, 25 Mar 2004 17:52:00 GMT

Bill announcing the launch of Microsoft Speech Server 2004

Richard demonstrating the Prompt Editor tool in the Speech Application SDK

Richard calling up the application he's just authored live. Tense moments...

One of the Microsoft 'Speech Girls'

Speech Server 2004 wins the Visual Studio Magazine “Editors Choice Award“

David Burd with just a small portion of the Speech Server balloons we've got around the conference

Transcription and slides from Bill Gates keynote at VSLive/SpeechTek

johnlawr — Thu, 25 Mar 2004 15:24:00 GMT

Complete transcription of Bill's speech - including a transcription of all the demos and the accompanying powerpoint slides are here. The Speech Server section is the last 1/3 of the speech - just search for “Speech Server“

http://www.microsoft.com/billgates/speeches/2004/03-24-VSLive.asp

Bill Gates' Speech Server Keynote is online

johnlawr — Thu, 25 Mar 2004 07:21:00 GMT

Richard just sent me this link to the video of Bill Gates' keynote speech including the launch and demo of Microsoft Speech Server 2004.

http://www.ftponline.com/reports/vslivesf/2004/gates/

Some photos from the Speech Server Launch

johnlawr — Thu, 25 Mar 2004 00:00:00 GMT

Richard Irving demonstrating the grammar editor

Richard showing off the richer (and complex) power of the speech DataTableNavigator control

Richard getting animated about the extremely “sexy“ tools he is talking about!

Microsoft Speech Server has Launched!

johnlawr — Wed, 24 Mar 2004 23:43:00 GMT

Speech Server 2004 has been launched!

After 5 years of working towards this day I can say that it's been amazing to see everything finally coming together so well. The demos worked flawlessly - the keynote speeches from both Bill and Kai Fu were extremely well received, and the interest we're seeing on our booths and on partners booths has been fantastic.

For me personally, probably the highlight was a private demo we gave to Bill on the stand before the doors opened. We demonstrated the New York Department of Education application which has been authored by Intervoice where parents can phone up and find out their child's attendence records etc. Despite one false start due to excessive background noise we were able to show it working pretty well.

I've spoken to a number of conference attendees today - I think we're finally seeing Microsoft becoming accepted as a major player in the Speech industry, and I've been really thrilled by the number of people who have been in the industry for years and are really looking forward to having a standard non-proprietary platform they can work against.

Major thanks to everyone who has put this all together - I'm just disappointed that not everyone on the team has been able to come. Huge thanks to everyone who has done their part on my team to make this vision a reality - Anthony Bowker, Dave Pickford, Harry Morgan, Mark Cecys, Nilkund Aseef, Dave Whitehouse, Joe Marin, Mike Miller, Meredith Shaw. I wish you could all be here. Thanks.

New log analysis tools in Microsoft Speech Application SDK

johnlawr — Mon, 22 Mar 2004 16:44:00 GMT

As I mentioned in my last post, we've been working on some tools to analyze and report against the copious logs that are generated by the Microsoft Speech Server.

What started as a bit of a stretch goal for us to at least have something in the product to be able to look at these log files has turned into a full blown effort, and we've ended up with some pretty rich functionality (although there is always more we could to in vNext).

The MSS logs are generated on the server using Enterprise Instrumentation Framework (EIF). This has the advantage of being very low impact on the server runtime - the performance hit on even logging vast quantities of events is very low. It's also highly flexible in terms of being able to configure what events get added to the logs, and it supports multiple different log syncs.

However, for log analysis tools this makes for a different story. In order to reduce runtime overhead, the logs are stored in a binary format which is great for writing to sequentially, but for the kind of random access queries we'd like to be able to perform on the logs, it is less than ideal. Additionally, the flexibility of the format in terms of configuration and the complexity of the schema used for all the different types of events that can get logged by the server has made it hard to design robust tools with which to analyze the logs.

In order to do anything useful we decided that we'd have to get the log data into a more appropriate format for querying and reporting against, so we've put together a command line tool and parallel Data Transformation Services task to import the data into a database. On top of this we have two main components:

CallViewer: a standalone tool to run queries against the speech server logs. This tool can be used to find calls matching a specific set of criteria, and then to see the events you're most interested in within those calls. A typical use would be to see summaries of all of the Question & Answer cycles within a call. It's easy to get a quick overview of the 'shape' of an individual call - what the system said, how the user responded etc., and then to actually hear a recording of what the user actually said to the system.
Speech Application Reports: A set of standard reports designed to display an overview of Microsoft Speech Server usage and how well specific applications being hosted on the server are performing. We've developed this on top of the newly launched SQL Server Reporting Services. (I believe we may be the first Microsoft server offering shipping with reporting built on top of this framework). This is a pretty extensible framework, and our goal has been to provide a baseline set of reports which would be easy to augment with custom reports by customers and ISVs.

I'm hoping that some of our partners in the speech industry would see this as a real opportunity to add value to the MSS community. Tuning of speech applications during test and pilot stages is one of the most important phases of speech application development, so providing further reports to help application authors determine what users are struggling with, or which parts of their application need the most work could be very valuable indeed.

However, I'm pretty happy that we've plugged a huge hole in our end-to-end development story by providing these tools out of the box with the Speech Application SDK.

2 days to Microsoft Speech Server launch event at SpeechTek

johnlawr — Mon, 22 Mar 2004 16:06:00 GMT

It's been a long time since I posted. My excuse is that we've been in crunch mode working hard to get everything ready for the launch this week.

At one point in the product cycle I had mistakenly thought that things were well under control and that we'd have no trouble hitting our target dates. However, this was before we decided to take on another major new set of functionality for the Speech Application SDK to generate reports and analyzing data from Microsoft Speech Server logs (which I'll talk about elsewhere).

This has been combined with the typical madness that leads up to any sort of product lauch (final last minute discovery of bugs, trying to make sure that the demo that is going to be shown on stage when Bill speaks is water tight, and a vast number of fit and finish issues). Consequently I'm feeling rather frazzled and haven't really had time to compose my thoughts for long enough to write them down here.

However, tomorrow I'm heading to SF and it's all starting to get exciting hearing about the plans. We're expecting something like 4000 people hear Bill speak at the keynote on Wednesday. Apparently we're supplying donuts for the event - we've ordered in excess of 6000 Krispy Kremes. After Bill's keynote there will be a further keynote for the SpeechTek audience in the Metreon cinema by Kai-Fu where we'll be showcasing some of the applications developed by our partners and early adopters.

Tomorrow it will be exactly 5 years since I started heading up a team of developers to produce a set of speech application authoring tools at a small company called Entropic. When I think of the fact that 5 years later these tools are going to be launched in front of 4000 developers by Bill Gates... well it blows me away.

See you at SpeechTek/VSLive!