A Tutorial Introduction to SALT2. Software InstallationAlthough commercial implementations of SALT are becoming available, we will demonstrate SALT using the free Microsoft Speech add-in for Internet Explorer. This product is limited to US English, but is satisfactory for small English language applications. 2.1 System RequirementsYou will need at least the following hardware for speech applications development:
You will need the following software for speech system development:
Use Windows Update to install these and all recent patches prior to installation. You should also have some suitable means for editing and testing web pages. The examples in this tutorial have been developed using the free 1st Page 2000 editor from EvrSoft. 2.2 Installing Speech Add-inThe Microsoft Speech Add-in is normally distributed as part of the Microsoft Speech Application Softare Development Kit (SASDK). However this kit is over 300MB in size and contains much more stuff than we need for this tutorial (it contains all the server-side tools too). In section 2.3 below you will find instructions about how to download the full SASDK. However, a number of people have extracted the Speech add-in from the SASDK and packaged it for separate installation. This reduces the download to (only!) 65MB. Here are some places where you can download the add-in by itself:
You will need to be logged in with Adminstrator privileges to install the add-in. 2.3 Installing full Speech Application Software Development KitHere are the instructions for downloading the full Speech Applications SDK from Microsoft. Once you have downloaded the full kit, you still need to decide whether to install all components or only the Speech add-in for IE. If you want to install the complete SADK, you will need this software in addition to the software described above for the add-in:
To download and install the SASDK, go to: To download the Speech Application SDK in one step, download the file SASDK_V1_Full.exe and save it on your computer in a temporary folder. This file is over 300MB, so do this on a machine with a fast network connection! Create a directory c:\SpeechSDK and run the file SASDK_V1_Full.exe specifying the destination directory as c:\SpeechSDK. The unpacked files will occupy at least 1GB. You can now delete "SASDK_V1_Full.exe". Now download the file SASDK_Extract.exe and save to c:\SpeechSDK. Then run the file to extract the installation script "SASDK_ExtractAll.cmd" into c:\SpeechSDK. Start up a Command Prompt window and type C:\ cd \SpeechSDK C:\SpeechSDK> SASDK_ExtractAll.cmd C:\SpeechSDK\Setup You can now delete the *.CAB files in c:\SpeechSDK. If you want to install the full SASDK, browse to c:\SpeechSDK\Setup and run Setup.exe to actually install the software. You can read the file C:\SpeechSDK\Setup\Readme.htm for full instructions on installing the Speech Application SDK. If you just want to install the Speech add-in, browse to c:\SpeechSDK\Setup\Redistributable Installers\Speech Add-in for Microsoft Internet Explorer and run Setup.exe. Unfortunately the setup program does not install any SALT documentation, so you will need to use the links on the front page of this tutorial for help. You can now delete c:\SpeechSDK. 2.4 ConfigurationYou must train the speech recogniser on your own voice to get acceptable recognition performance. To set up the speech recognition and synthesis tools, find the Speech applet in the Control Panel. The Speech Recognition tab allows you to add users by name and to create a personal voice profile for each. You can also use this applet to test and configure your microphone. The Text to Speech tab allows you to set the default synthesizer voice and talking speed. Next: a first web page.
|
|
University College London - Gower Street - London - WC1E 6BT - |