(FEB 2008) — Adding another name to the the long list of voice platforms on which LumenVox is certified, LumenVox announced this week that our Speech Engine has been rated compliant by Avaya.
After considerable testing by Avaya, the LumenVox Speech Engine was certified as compliant with Avaya Voice Portal 4.0 and Interactive Response 3.0. The tests were performed using Media Resource Control Protocol (MRCP) version 1 as a method of communication between the Avaya platforms and the Speech Engine.
In addition to offering Avaya customers another choice of speech engines, the integration reflects the general maturity of MRCP. As part of our commitment to standards–based speech recognition, LumenVox supports LumenVox Media Server as a method of communicating with our Speech Engine.
Featuring a very slick graphical interface to manage the names/extensions and perform tasks such as adding in custom pronunciations and nicknames, the company directory is a free, open source application that is ready to be downloaded now.
For the full source code and a demo, see Ethan's FreePBX post on the directory.
The other application turns a button on a phone into a "magic button" that causes the PBX to play a tone and then allows a user to issue a command by speaking it. These commands work whether a call is in progress or not.
This means a user can pick up the phone, press the button, and say "Call John Smith" to dial that person. The same user can, during the call, press the button and say "Park Call" to park the call or perform a variety of other functions.
Ethan will be demonstrating the application and giving away all of the source code at the FreePBX training class at the end of the month, in the meantime he has another post with more information about the application.
Also check out the Feb. 15 VoIP Users Conference Call to hear the applications in action.
As part of our ongoing efforts to innovative speech recognition technologies, we are happy to announce that we have recently been awarded two new patents on key technologies used by the LumenVox software.
The first is for a method of determining confidence scores by creating an array of time frame and acoustic scores (the raw score from the acoustic model) and determining the phoneme sequence with the highest score under certain constraints.
The second patent is for a method of recognizing words in a natural language type of environment, using the context of several keywords in order to figure out their semantic interpretation.
We are currently building a section on LumenVox.com to promote some of the ways our customers have used our technology. This area will feature recordings of actual calls into the applications, providing a technology demonstration and a short case study.
This section of LumenVox.com will also have a brief description of the application and a link back to the developing company's Web site, providing good exposure for the application and the company responsible for it.
To participate, all you need to do is provide us with a brief audio file (approximately 1–5 minutes long) of somebody calling the speech portion of your application.
If you need assistance with this, we're happy to help you out by setting up a conference call with you and recording your interaction with the application.
For more information, please email us. You may also contact us by phone, toll free in the United States at 1–877–977–0707, or internationally at +1–858–707–7700. Just ask to talk to "Sales."
Our next release, tentatively scheduled for March, will include some relatively major changes for our Linux users.
As anyone who has installed our software on Linux knows, our current packages install everything into the directory /opt/lumenvox/ and set up a number of environment variables that point to specific subdirectories.
This causes problems in the event that these variables are not set — a particular problem when using scripts that have different shells to start our software. Starting with the next release, our Linux packages will install into different directories, depending on the file type.
For instance, all of our executables will be placed in the standard /usr/bin/ and libraries in /usr/lib/.
We will no longer rely on environment variables at all, instead providing Linux users with configuration files if they wish to relocate some of our files.
All of this should make running and installing our Linux software considerably easier. We'll have full details when the release is ready, but be aware if you're planning on installing or upgrading LumenVox software on Linux later this year that things will be different.
© 2016 LumenVox, LLC. All rights reserved.