I spent much more time evaluating this product than I had expected. My conclusion: it might be a good solution for those who build mobile applications for Windows CE with Microsoft Embedded Visual Studio C++ as there is only C/C++ interface in Loquendo API. For those who develop in C# with Microsoft Visual Studio 2003 or 2005 is important to know that neither wrapper assemblies for ASR nor for TTS are supplied. So there is no direct possibility to utilize the required libraries in a C# (or VB.NET) project. I have to admit that was no-go criteria for me.
I could deploy the sample application for ASR though and it made a very good impression in terms of speech recognition quality and performance. Moreover the ASR is speaker independent (no training needed for every specific user), based on neural networks and Markov Models technologies and supports many established standards for grammar definition. Unfortunately I had less luck with the TTS sample application. The TTS runtime was properly installed on the Pocket PC, but the sample application still complained about some initialization failure directly on the start.
And the last important remark, the runtime must be licensed and activated on every PDA where it is installed. It might make the roll out process a bit more complicated.
Here are the links for further information:
And the last important remark, the runtime must be licensed and activated on every PDA where it is installed. It might make the roll out process a bit more complicated.
Here are the links for further information:
8 comments:
thanks, very useful information !!! How's going on with Nuance/Vocon?
Thanks for your feedback. I think I'll have some more infos next week.
Loquendo has a very nice set of c# wrappers for their speaker ID, language ID etc products. Might be worth asking them again, as these might be new.
YMMV
Paul
Hi Paul,
thanks for your info. We asked Loquendo for .NET wrappers during the evaluation time. They provided us some classes for Win32 platform, which unfortunately did not work under Windows Mobile. I suppose they were at a very early development stage (or libraries for Win32 and WinCE are so much different). In the meantime we are pursuing a solution with Vocon, a product of Nuance and are very satisfied with it as well as with provided support. A pilot application is almost completed. Recognition results and performance are surprisingly good even for such a skeptic as myself.
Kind regards, Andriy
Hi,
is there any follow-up on Loquendo and/ore Nuance speech technologies for mobile?
Almost a year has passed since the post was created and I am eager to find out what is state-of-the-art now.
Thanks
Hi Dusan,
unfortunately I don't have further information on Loquendo products. ASR/TTS is a very complex topic and we are glad, we have now reached comparatively good results in binding Nuance speech engines in our applications. So Loquendo is not pursued in the meantime, but they would have been the next possible alternative if Nuance were a failure. As Paul wrote, Loquendo may presently have C# wrappers as well as provide better support. Try to visit their web site.
I am interested in developing a multi modal UI and I would certainly need dynamic grammar generation. Does Nuance offer it?
Is it actually VoCon mobile XG that you are referring to?
Do they have eval version of the SDK?
Thanks
the product of Nuance we are using is called VoCon 3200 SDK. It does support dynamic grammar compilation, but it is a very advanced topic.
There is no evaluation version to download, but they can provide it on request. I would suggest consider to buy a couple of consulting days from them. It helps to save plenty of time at the beginning stage to get understanding of speech recognition and how the SDK works and can be used.
Post a Comment