Previous Topic

Next Topic

Book Contents

Book Index

Home Page

How to Change Pronunciation for Text-to-Speech and Speech Recognition

Active Call Center includes a utility that allows you to modify the pronunciations used by the text-to-speech engine and the speech recognition engine. This utility may be helpful for building custom vocabularies.

The features provided by the utility are actually provided by Microsoft's Speech API and we have found a variety of limitations in its functionality. We do not encourage use of this utility due to these limitations, however, we recognize that some customers will find the utility a useful, if not frustrating, way to customize the speech engines.

Since the features of this utility are provided directly through the Microsoft Speech API, we have no way of fixing bugs that may exist and no way to provide advanced technical support regarding its use. We recommend that only users who absolutely must have customized speech recognition or text-to-speech pronunciation (for example, foreign customers) use this feature.

Change Text-to-Speech Pronunciation

To change pronunciations for text-to-speech, follow these instructions:

Step 1. From the Active Call Center menu bar, select Tools ... Pronunciation Editor.

Step 2. Select a text-to-speech engine that will be used from the list presented (the list includes speech recognition engines also). Pronunciations cannot be modified for the Lucent text-to-speech engine. Other engines may also have vocabularies that cannot be modified with this utility (for example, IBM ViaVoice Outloud engines did not work properly with this utility).

Step 3. A dialog window titled Lexicon will appear. Words may be edited one at a time by clicking Edit Another. Follow the instructions that appear in the wizard after clicking Edit Another. If the word pronunciation gets messed up from editing it, remove the pronunciation by clicking Edit Another again, typing the word, then selecting Remove this word from the dictionary.

Step 4. Words may also be loaded in a batch from a text file. To load words in this fashion, click Scan File for Words and follow the prompts. Once all the words are loaded, they may be edited one at a time and added or removed from the text-to-speech engine's dictionary.

Change Pronunciations used by Speech Recognition

To change pronunciations for speech recognition, follow these instructions:

Step 1. From the Active Call Center menu bar, select Tools ... Pronunciation Editor.

Step 2. Select a speech recognition engine that will be used from the list presented (the list includes text-to-speech engines also). We have only tested this utility with the Microsoft Speech Recognition Engine.

Step 3. A dialog window titled Lexicon will appear. Words may be edited one at a time by clicking Edit Another. Follow the instructions that appear in the wizard after clicking Edit Another. If the word pronunciation gets messed up from editing it, remove the pronunciation by clicking Edit Another again, typing the word, then selecting Remove this word from the dictionary.

Step 4. Words may also be loaded in a batch from a text file. To load words in this fashion, click Scan File for Words and follow the prompts. Once all the words are loaded, they may be edited one at a time and added or removed from the speech recognition engine's vocabulary.

Step 5. We have found that it is not possible to save multiple pronunciations for the speech recognition engine. In fact, the utility will actually delete multiple pronunciations that may be present in the speech recognition engine's vocabulary. If this happens, reset the pronunciations for that word: edit the word individually by clicking Edit Another, type the word, and then select Remove this word from the dictionary.

Step 6. If there is a need to reset the speech recognition engine's vocabulary, this can be accomplished by uninstalling and reinstalling the speech engine as follows: From the Windows Taskbar, choose Start ... Settings ... Control Panel ... Add/Remove Programs ... Microsoft Speech Recognition Engine. Then open the Active Call Center program folder in Windows Explorer and run the application mscsrgphl.exe - this will reinstall the Microsoft Speech Recognition Engine.

See Also

Advanced Audio Topics

How to Play a Sound File to the Speakers During a Call

How to Increase Volume of a Wave File through Scripting

How to Change Recording Silence Detection Levels

How to Enable Fast Forward/Rewind and Other Special Playback Features

How to Obtain Foreign Language and Other Speech Engines

How to Join Multiple Wave Files into One Wave File

How to Extract Sections of a Wave File

How to Analyze and Compare Recorded Audio Data