Welcome to the TI Embedded Speech Recognizer (TIesr) Home Page
The TI Embedded Speech Recognizer (TIesr) is a fixed-point recognizer written in C++ and C. It is designed to balance resource usage, robustness to environment, and performance. It has a simple easy-to-use interface. This makes it an excellent recognizer for developing a wide variety of voice-enabled embedded applications. TIesr is a medium-sized phonetic-based recognizer that can accommodate a vocabulary of up to several hundred words. The words and phrases to be recognized can be dynamically changed by an application. The TIesr project release currently comes with general English language support, and tools to develop the support for other languages. TIesr has been tested on Linux OS platforms, including the Beagle Board and several OMAP EVM boards (35x, L138, Zoom) as well as platforms running Windows and Windows Mobile.
Note: TIesr software is now licensed under the LGPL v2.1.
In order to join the project, you must have a registered account on gforge.ti.com. Click the link “Register new account” on the upper left corner. This will take you to a registration page. Fill in the form and get registered as a gforge.ti.com user. Once you are registered as a user, log in to the gforge.ti.com web site by clicking “Log in”, then visit the TIesr project summary page and click “Request to join project”. Once the request has been processed and you have been added to the project as a developer, which may take a while, you will have access to the TIesr project subversion repository.
Approval has been granted to offer the TIesr Embedded Speech Reocognizer library as open source under the LGPL v2.1 license. The latest version, Beta_1.3, reflects the open source licensing.
As of March 10, 2010, the TI Embedded Speech Recognizer project is available.