Speech Technology Magazine

 

Chris Lotspeich, Director of Marketing, LumenVox

Chris Lotspeich, director of marketing at LumenVox, explains that "Other tuning tools typically look at the enterprise level of the application, whereas the Speech Tuner gives an in-depth look 'under the hood.'"
Posted Jul 1, 2005
Page1 of 1
Bookmark and Share

Q. Chris, what is going on with LumenVox?
A. Well, work on our core Speech Recognition Engine has been focused around supporting MRCP, n-best, and porting the software to Linux.  The Speech Driven Assistant for TeleVantage is in version 2.0, with improved support for outbound calling to user's contacts and email maintenance.  About a month ago, we announced the availability Nuance 8.5 within our Speech Tuner.  At SpeechTEK, we will be announcing beta support for ScanSoft OSR 2.0.  The Speech Tuner allows companies working with Nuance, ScanSoft, or LumenVox speech applications to perform tuning and testing in-house.      

Q. Speaking of the LumenVox Speech Tuner, please elaborate a little more on this product.
A. LumenVox's Speech Tuner is a complete maintenance tool that allows end-users, value-added resellers, and platform providers to transcribe their call data, and then tune the application by working with engine parameters, application grammars, or even version upgrades of any speech-driven application created on Nuance 8.5, ScanSoft OSR, and LumenVox ASR engines.  The Tuner allows companies to tune and test each interaction immediately, providing highly dynamic interactive tuning and testing sessions.  Using their own call logs, companies can get a detailed view of how the ASR is responding to callers.  They can make changes to grammars and engine parameters, and then immediately retest against historical data to determine if any change would help, hurt, or make no difference to the application's performance in the future.

Q. What benefits can your customers gain from this solution?
A. Customers gain much more insight into how their speech applications are performing on a call-by-call basis.  Other tuning tools typically look at the enterprise level of the application, whereas the Speech Tuner gives an in-depth look "under the hood."  But, of course, our Tuner also has many great higher-level reporting features as well.

Q. How in-depth is the testing that the Tuner provides?
A. The Tuner exposes every detail of the ASR engine call logs, from barge-in information and ASR results all the way up to application-specific log information.   That information can be used to adjust every available ASR engine parameter, as well as full testing with grammars, including semantic interpretation.  The Tuner automatically provides statistical information during each test, to help evaluate performance changes.

Q. Why did you decide to add functionality to the Speech Tuner for the other ASR platforms?
A. When we first introduced the Speech Tuner at SpeechTEK 2003, the number one question was, "When will you support the other speech vendors?"  A little bit of market research later, we figured out most people were looking for Nuance and ScanSoft support, closely followed by support for the other speech players.  The decision to move forward has largely been the call of the customers that we have met and spoken with.  Many current features in the Tuner are the direct result of customer requests.

Q. Will there be additional speech vendor support in the near future?
A. After we officially release the support for ScanSoft OSR 2.0, we are planning on incorporating Microsoft's Speech Server, and other industry ASRs.  We are also looking at supporting several VoiceXML platforms including Vocalocity, SandCherry, Voxeo, and Vocomo.

Q. Any last thoughts?
A. Yes, we will be demonstrating the Speech Tuner and how it works at the upcoming SpeechTEK show in New York.  We are going to be offering these demonstrations every hour on the hour in booth 804.  We will also present our core Speech Recognition Engine for people who have yet to see the power and flexibility our ASR offers.  These presentations will run on the half-hour of every hour.  Each presentation will take about 10 - 15 minutes and we encourage everyone to stop by our booth and "Discover LumenVox."

Page1 of 1
Learn more about the companies mentioned in this article in the Speech Technology Buyer's Guide:
Learn more about the companies mentioned in this article in the Vertical Markets Guide: