These Release Notes apply to SpeakToText Version 2.51 dated December 14, 2006 and released on December 22, 2006. Following these release notes is a section on Known Issues, which may also be helpful. Additional release notes will be posted on our website (www.CoolSoftLLC.com) as they become available.
This Version 2.51 fixes a number of bugs found in Version 2.50, including:
· A crash upon entering settings if the microphone was turned off.
· Missing tool tips for user commands in the What Can I Say? dialog.
· An error in processing User Commands if the path contained spaces.
· A bug that prevented the Target field in User Commands from being edited.
This release also includes an important enhancement to User Commands. User Commands can now process commands that contain command syntax for an application. The vertical slash "|" is used to separate the application path and the command syntax.
To create a User Command that runs an application and gives it a command, the entry in the Target field of the Create User Command dialog has two parts: (Application Path|Command Syntax). The first part before the vertical slash must contain the path to the application, and the second part after the vertical slash contains the command to be processed by the application when it runs.
Typically, any “Application Path|Command Syntax” combination that runs from the command line, will also run in User Commands by speech. Take the following example for turning on Module A1 in an X10 Active Home Pro home automation system.
C:\Program Files\Common Files\X10\Common\ahcmd.exe|sendplc a1 on
Release 2.50 – Dated September 11, 2006 – Released on September 30, 2006
SpeakToText is now titled “SpeakToText 2.0 Speech Recognition” in order to more accurately represent its functionality.
New Features in SpeakToText™ 2.0 Speech Recognition, Release 2.50:
Speech Recognition
Additional enhancements to the speech recognition layer to make speech recognition more robust.
Added speech recognition engine management between CoolSoft products. If more than one CoolSoft speech recognition product is running, and the microphone is turned on in one of them, the other turns off its microphone and releases the speech recognition engine.
Improved Microphone Management. The speech recognition engine is released when the microphone is turned off. Added a red/green light to the lower right to indicate when the microphone is on and SpeakToText has control of the speech recognition engine. Microphone is turned off and the speech engine is released when the microphone button is clicked or the microphone off command is spoken, after clearing the speech buffer, when going into speech properties using the speech properties button or speech command, and when the microphone is turned on in another CoolSoft speech recognition product. The microphone is disabled during typing and reading without releasing the speech engine.
Improved .wav file processing. SpeakToText no longer minimizes and now lets you see the progress as a .wav file is being processed. Allows use of Clear Speech Buffer to abort .wav file processing. Microphone is turned back on after .wav file processing is complete.
Text to Speech
Improvements in Text to Speech. Modified Read Text button and speech command. Added pause and stop reading buttons.
Send Message Feature
Added option to auto-send chat messages every 30 seconds
User Commands
Custom Commands renamed to User Commands. Improvements to User Commands. Added “sleep” and “focus” actions, and improved management of multi-step user commands. User commands section added to What Can I Say? dialog.
New Speech Commands
Added “Tab Key” speech command to send the “tab” character.
Added Scroll Down and Scroll Up speech commands, which act on SpeakToText in basic mode, and on the What Can I Say? dialog when it is open.
Added launch CoolInfo speech command to launch CoolInfo if it is installed.
What Can I Say?
Improved sorting and management of What Can I Say? dialog. Added tool tips to What Can I Say? commands list.
Registration
Registration of a license key is now verified on-line and limited to five times, in accordance with the user license agreement. Must be connected to the Internet to register.
Added sponsor management to registration process. The registration process is now capable of giving credit to sponsors if a copy of SpeakToText comes from an approved sponsor.
Updates
Updates from previous versions also update the Manual and Quick Start Guide to the most current version 2.50. When new updates or additional release notes become available, the information will be posted in the Support section of our website at: www.coolsoftllc.com.
Documentation
Documentation in the Manual and Quick Start Guide has been updated and improved.
A Quick Reference Guide is now available at http://www.coolsoftlllc.com/downloads/SpeakToText2_0QuickReference.pdf
Release Notes and Quick Reference Guide now included in Appendix of User Manual.
Release 2.35 – Dated October 14, 2005 – Released on October 14, 2005
This latest release of SpeakToText™ 2.0 includes two major upgrades to the speech recognition layer, and the addition of three major features to enhance usability, enable users to create their own speech commands, and read text. It also includes other new features, and numerous bug fixes.
New Features in SpeakToText™ 2.0, Release 2.35:
Dynamic Grammars
SpeakToText is now enabled to use dynamic grammars. This makes SpeakToText more robust, and will enable more features in future updates. The change is user transparent.
Re-entrant .wav File Processing
This change is user transparent, and has improved performance.
Advanced Mode
Set Target and Echo to Target functionality have been replaced by Advanced Mode. The target application no longer needs to be set, and when in Advanced Mode, you can change from target to target by clicking with your mouse in the desired target application. The desired target application MUST have the active focus, and the cursor MUST be placed in the text area of the desired target application where you would normally type your text. The copy to target functionality is no longer necessary and has been eliminated. You can still copy text from SpeakToText to other applications by way of the clipboard.
Custom Commands
SpeakToText now enables the user to create their own speech commands. After a Custom Command is created, SpeakToText will run the command whenever it is spoken. To create custom commands, see the Custom Commands tab in the Settings dialog.
Text-to-Speech (TTS)
A TTS feature has been added. SpeakToText will now read text from SpeakToText, the clipboard or the target application in advanced mode. The speech commands “read all” “read selection” “read right” “read clipboard” and “read target” (advanced mode) have been added. When SpeakToText is reading, the microphone is turned off. This is so that SpeakToText cannot hear the text that is being read. A “Stop Reading” button has been added. Clicking on the “Stop Reading” stops the reading.
Click Minimize
A “Click Minimize” speech command has been added to compliment the “Click Focus” command. Saying “click minimize” will minimize SpeakToText to the taskbar.
“Copy Target” Speech Command
This speech command copies the entire contents of the target application to the clipboard.
Bug Fixes
Many user transparent and minor bugs have been fixed. Major bug fixes include:
The “What Can I Say?” dialog is now empty when the microphone is turned off, and the start/stop listening buttons are unavailable.
The “Apply” button is now enabled in all tabs of the Settings dialog after making changes.
Loss of focus after copying to Microsoft Word when Word is started by SpeakToText is fixed.
Running more than one copy of SpeakToText is now disabled. When SpeakToText starts, it now checks to see if it is already running and prevents a second copy from being started.
Unresponsiveness during updates is fixed. The update dialog shows the progress of the download, and can now be canceled. Canceling the update restores SpeakToText.
A problem that caused the first update to be repeated in subsequent updates has been corrected. This problem occurred only on some systems. If you are experiencing this problem, go to the SpeakToText v2 folder in Program Files, delete the SpeakToTextSetup.exe file, and run the update again.
Updates
Updates from previous versions also update the Manual and Quick Start Guide to the most current version 2.35. When new updates or additional release notes become available, the information will be posted in the Support section of our website at: www.coolsoftllc.com.
Documentation
Documentation in the Manual and Quick Start Guide has been improved.
Additional Changes and Known Issues Determined After Completion of the Manual
If you are installing the full version of SpeakToText after a previous install of the trial version, or if you are reinstalling SpeakToText without having uninstalled it first, you may get an error message that says: “The file 'C:\Program Files\SpeakToText v2\UNWISE.EXE' could not be opened…” If this happens, exit the install, and start the install over. This second time you start the install, it should run through to completion. If not, then submit a technical support request in the support section of our website.
If you are using the SAPI 6.1 Speech Recognition Engine in Windows XP systems, background noise filtering can be sensitive on some systems. While the SAPI 6.1 Speech Recognition Engine will give you better performance in this configuration, enabling background noise filtering for dictation and/or commands in SpeakToText may result in the loss of dictation and command functionality if filtering is too strict. After you enable background noise filtering, if you lose speech command functionality or dictation appears in SpeakToText then disappears, use your mouse to return to the Settings dialog and start moving the sliders under the two filtering options to the left until command and dictation functionality returns. This is a matter of trial and error to find what levels work best on your system.
When starting SpeakToText, if your speakers are turned up too loud when the greeting is spoken, or if there is a high level of background noise, a few stray words may appear in SpeakToText while the speech recognition engine is initializing. If this happens, you can simply clear the unwanted text with the “clear text” “scratch all” speech commands, or by clicking on the “Clear SpeakToText” button. You can also turn down your speakers or eliminate/reduce the background noise to prevent this.
If you want your speech recognition profile to learn from corrections you make to your dictation, your corrections must be done within SpeakToText, and must be done by selecting alternate words or series of words from the right click menu. Corrections typed into SpeakToText, or made in a target application while in Advanced Mode, will not provide training to your speech recognition profile. To make corrections from the right click menu, right click on a word, or highlight a short series of words and right click to display alternates for that word or series of words. Selecting one of the alternates (if there is one) will replace the word or words in SpeakToText, and will have the effect of training your speech recognition profile. Note that not all words have alternates. Alternates are displayed only when the speech recognition engine detected more than one possible word or phrase with a high enough score to be probable. Also note that “Background Adaptation” must be enabled for your speech recognition profile to learn from corrections made in the right click menu. To enable Background Adaptation, go to Speech Properties, select your recognition profile, click the settings button and there is a check box at the bottom to control Background Adaptation.
The registration dialog has been modified so the entire registration key can be seen when it is entered.
Release 2.01a – April 12, 2005
This release fixes problems with registration keys that did not work with some combinations of First name, Last Name and Email Address.
This release incorporates text in the General Tab of settings for an as yet unreleased feature. The text and checkbox, “Use Custom Commands” is grayed out and has no effect. When finished, the Custom Commands feature will be included in an update, so please check our support pages periodically for announcements about this exciting new feature.
The functionality is unchanged from Release 2.01, and the same Manual and Quick Start Guide apply to release 2.01a. If you did not have any problems with your registration key, then you do not need to upgrade.
Release 2.01 – February 28, 2005
The SpeakToText Install Takes a Long Time
This is normal on many systems. First of all, the SAPI 5.1 speech engine is a large file that takes time to unpackage and setup. Second, if you are running anti-virus software, it may be scanning some of the files being installed, and this can slow down the process. Third, installs are particularly slow on older systems (with slower processors or less than 256 megs of memory). Combine these three factors together on one system, and the install can take a long time (over five minutes).
If the install is taking a long time, please be patient and wait for the install to complete. Chances are that nothing is wrong and the install will finish normally. If you stop and start over again, it will just end up taking longer, so do not cancel and restart the install unless you are certain the system has stopped responding.
Filtering Commands for Background Noise
There was an issue in Release 2.00 where sometimes background noise was recognized as a command. For example, when listening was off and SpeakToText was not being used, on some systems “What Can I Say” would open or listening would become enabled when the command was not spoken, leading to other commands being recognized and processed when nothing was spoken. This happened because the SAPI speech engine first tries to interpret unidentifiable background noise as commands. Turning off the microphone when not using SpeakToText solved this issue, but we decided to also add filtering of commands for background noise.
Most users will not need to enable this feature, but if you are having a problem with background noise being recognized as commands, you should enable this setting. In settings under the formatting tab, a slider bar has been added to filter commands for background noise. The default setting is “off.” To enable the feature, check the box. You will need to determine the optimal level of filtering by trial and error, as the range of scores assigned to commands varies widely from system to system. Too much filtering will result in spoken commands being rejected. Too little will mean background noise continues to be recognized as commands, which are then processed. More detail is available on this feature in the “Background Noise Filtering” section of the manual, and we recommend reading that section if background noise is a problem for you.
Issue with Outlook 2003 Resolved
There was a issue in Release 2.00 with Outlook 2003, where if you closed an email compose window, SpeakToText did not release Outlook as the target, and when you went back to Outlook to create another message, text dictated after your last message was copied to the new message. This issue is resolved in release 2.01.
Listening Information Dialog Box Issue Resolved
In Release 2.00, the Listening Information Dialog Box that explains the “Listening” feature at startup was sometimes getting lost in the background if the user was working with multiple programs when they launched SpeakToText™ 2.0. This gave the appearance that SpeakToText had not started, even though that dialog was in the background waiting for the user to click on it. The Listening Information Dialog Box displayed upon startup now times out, enabling SpeakToText to start even if this dialog gets lost in the background.
Copy To Word Issues Resolved
Issues in Release 2.0 that prevented copying to Word 2000 on some systems has been fixed.
Click Focus Speech Command Added
A “Click Focus” command has been added. If SpeakToText has been minimized or moved to the background after copying text or using another application, then provided the microphone is on, saying “click focus” will return SpeakToText to view and the program will have the focus. This “click focus” command is available with listening in the off position, but like all other speech commands, it is not operational if the microphone is turned off.
Updates
The update now launches SpeakToText when finished.
Release 2.00 – December 8, 2004
None
Release 1.95 – November 30, 2004
None
As of the December 22, 2006 release date, the following known issues apply to SpeakToText™ Speech Recognition Version 2.51. Many of these items are related to functionality that is normal behavior for SpeakToText, and the discussion gives recommendations for dealing with them. Other issues may be addressed in future releases, so we recommend that you check for updates periodically using the “Check for Updates” button in the ”About” dialog or the “Check for Updates” selection from the Help menu.
The SpeakToText Install Takes a Long Time
This is normal on many systems. First of all, the SAPI 5.1 speech engine is a large file that takes time to unpack and setup. Second, if you are running anti-virus software, it may be scanning some of the files being installed - this can slow down the process. Third, installs are particularly slow on older systems (with slower processors or less than 256 megs of memory). Combine these three factors together on one system, and the install can take a long time (over five minutes).
If the install is taking a long time, please be patient and wait for the install to complete. Chances are that nothing is wrong and the install will finish normally. If you stop and start over again, it will just end up taking longer, so do not cancel and restart the install unless you are cretins the system has stopped responding.
Install Failures
SpeakToText is installed with the Wise Installer. There are two known issues with the installer which are described below. If you encounter one of these problems and the recommended solution does not work for you, then submit a technical support request in the support section of the www.coolosftllc.com website.
If the install hangs, it is likely that another program has installed a copy of UNWISE.EXE in the root of the C:\ drive. The recommended workaround is to look in C:\ for the file “UNWISE.EXE. If you find it rename the file to UNWISE.OLD (so you can rename it back to .EXE if another program ever needs it), then try the install again.
If you are installing the full version of SpeakToText after a previous install of the trial version, or if you are reinstalling SpeakToText without having uninstalled it first, you may get an error message that says: “The file 'C:\Program Files\SpeakToText v2\UNWISE.EXE' could not be opened…” If this happens the recommended workaround is to exit the install, and start the install over. This second time you start the install, it should run through to completion.
Startup Related
Unwanted Text Appears When Starting SpeakToText
When starting SpeakToText, if your speakers are turned up too loud when the greeting is spoken, or if there is a high level of background noise, a few stray words may appear in SpeakToText while the speech recognition engine is initializing. If this happens, you can simply clear the unwanted text with the “clear text” “scratch all” speech commands, or by clicking on the “Clear SpeakToText” button. You can also turn down your speakers, use the speaker(s) on your headset while using SpeakToText, or adjust your environment to eliminate/reduce the background noise.
Misrecognition of Your Speech When SpeakToText First Starts
When you first start SpeakToText or after you turn the microphone on, recognition accuracy may not stable for the first 15 to 30 seconds. This is because it takes longer for the SAPI speech recognition engine to finish initializing and become stable, than it does for SpeakToText to startup. This happens at both startup and when you turn the microphone on because when the microphone is turned off, SpeakToText releases the speech recognition engine. When the microphone is turned on again, the speech recognition engine needs to reinitialize.
The Microsoft SAPI speech recognition engine is third-party software over which SpeakToText has limited control. After you begin dictation the speech recognition engine stabilizes, and the time this takes is dependent on your system. If after starting SpeakToText or turning on the microphone, the first sentence you dictate is misrecognized or incomplete, the solution is to use the “clear text” command and start your first sentence over.
Listening Starts by Itself
There is a known issue where if the microphone is turned on, and listening is turned off, background noise may be recognized as the “wake up” or “start listening” commands. This enables listening, and after listening is enabled background noise or speech are sometimes detected as dictation. If you are in Advanced Mode, unwanted text may appear in the program which has focus (the target application). The workarounds for this are to i) enable and adjust command filtering in the formatting tab of settings; ii) turn the microphone off if you are not planning to use SpeakToText for a while; or iii) if you are using another application, return to Basic Mode (“end advanced” speech command or button).
Crash After Default Speech Properties Warning Dialog
If you are using the default speech recognition profile, SpeakToText prompts you with a dialog reminding you to create and train a speech recognition profile, and offers you the option to go to Speech Properties to do this training. If you enter Speech Properties, the startup of SpeakToText is suspended, and resumes when you exit Speech Properties. This delay in the startup of SpeakToText may cause the program to crash. If this happens, it does not affect the training you did in Speech Properties. Restart SpeakToText and it will function normally.
The First Command or Words Spoken After a Pause in Dictation May Not Be Recognized
With Commands
After a pause in dictation, commands are not always recognized. The command may appear in the Language Bar, but not be acted upon. The recommended workaround is to repeat the command. You may also try enabling the Background Noise Filtering for commands feature in the Formatting tab under Settings.
With Dictation
After a pause in dictation, the first word(s) in a sentence is (are) missed or recognized incorrectly. The recommended workaround is to use the “scratch last” or “no no” command, and repeat your dictation. You may also want to make certain you are pronouncing the first word as clearly as the rest of your words.
Background Noise and Background Noise Filtering
On some systems and in some environments, there may be a problem with background noise being recognized as commands that were not spoken or unwanted dictation resulting in “stray words” that were not spoken appearing in the dictation area. SpeakToText can filter for background noise to prevent this, and if background noise is affecting dictation or commands, enable one or both of these options.
It is best to start with the default setting of the slider in the center position, however, background noise filtering will perform differently on every system, depending on many factors like processor, memory, sound card, microphone, system resources being used by other applications, your environment, and even your voice on that day. If the default setting doesn’t work try adjusting the level. Finding the optimal level for your system can only be determined by trial and error. If filtering is too high, then commands or dictation may be rejected. If filtering is too low, then you will continue to experience unwanted text from background noise.
If you are using the SAPI 6.1 Speech Recognition Engine in Windows XP systems, background noise filtering can be sensitive on some systems. While the SAPI 6.1 Speech Recognition Engine will give you better performance in this configuration, enabling background noise filtering for dictation and/or commands in SpeakToText may result in the loss of dictation and command functionality if filtering is too strict. After you enable background noise filtering, if you lose speech command functionality or dictation appears in SpeakToText then disappears, use your mouse to return to the Settings dialog and start moving the sliders under the two filtering options to the left until command and dictation functionality returns. This is a matter of trial and error to find what levels work best on your system.
Audio File Recognition Issues
WAV File Processing – Some Buttons Do Not Gray Out
When SpeakToText first starts processing a .wav file, all buttons except the Clear Speech Buffer, Save Text and Read Text buttons are grayed out. The Clear Speech Buffer button is active in case you want to abort the .wav file processing, and the Save Text and Read Text buttons are live (although not functional at this time) because SpeakToText has placed new text in the text area. We do not recommend clicking on the Save Text and Read Text buttons while a .wav file is being processed. Doing so will abort the .wav file recognition, and you will have to start over.
WAV File Processing in Trial Mode and the 1000 Character Trial Limitation
When in trial mode, SpeakToText has a 1000 character limit. If you process a .wav file from trial mode, when SpeakToText reaches this 1000 character limit, no more text will appear; however, SpeakToText must finish cycling through the .wav file before you can turn the microphone back on. Depending on the size of your .wav file, this may take some time. Please be patient.
Nothing Happens After Dragging an Audio File into SpeakToText
You drag & drop an audio file on SpeakToText™ 2.0, and nothing happens. It’s possible that the SpeakToText confirmation dialog that appears by default before processing an audio file is hidden behind My Computer (or Windows Explorer), and minimizing that should reveal the confirmation dialog. To prevent this from happening, it is best for My Computer/Windows Explorer not to fully overlap SpeakToText™ 2.0 when selecting an audio file.
After you become familiar with using SpeakToText™ 2.0 to process audio files, the best solution to this issue is to disable this confirmation dialog by checking “Do not display this prompt again.”
User Command Issues
Focus Command Step Does Not Always Work
If you create a multi-step user command, the command you speak must be the same for each step, and only the tool tip for the first command in the sequence will be displayed. You can use the move buttons in the User Commands Dialog to adjust the sequence of multi-step user commands. If one of the steps in a multi-step user command is Focus (the equivalent of the speech command “Click Focus”), this may cause a race condition where the Focus command wins the race and is executed before the other steps in the command. This gives the appearance that the focus command failed to execute, when in reality it was processed before the other commands in the sequence and SpeakToText was then covered by the other program(s) in the command sequence. To avoid this problem, the Focus step in a multi-step user command should be the last command in the sequence, and should be preceded by a “Sleep” command with a value that allows enough time for the other commands in the sequence to be processed before the Focus command is executed. For example, the command of “start my favorite programs” can execute the following commands:
Browse = www.cnn.com
Start = notepad.exe
Sleep = 10
Focus
This results in ww.cnn.com opening in the default browser, and Notepad starting, both of which will open on top of SpeakToText. The sleep command pauses for 10 seconds (allowing these programs to open), and then executes the Focus command which brings SpeakToText to the foreground, achieving the desired result.
The Command Step Following a Browse Command Step Does Not Always Work
If you are creating multiple step user commands, a sleep command needs to follow a browse command, or the command(s) that follow the browse command may not be executed. The time of sleep command following a browse command depends on speed of internet connection, and you need to experiment to find a sleep interval that works best for your system and connection speed. Note that a single step browse command, or a browse command that is the last command in a multiple step user command do not require a sleep command.
Background Noise and the Create User Commands Dialog
On some systems and in some environments, the Create User Commands dialog can be highly sensitive to background noise, causing it to close or change tabs after creating a User Command. If this happens on your system, simply turn the microphone off and user your mouse to navigate to the Create User Commands dialog (Settings button, User Commands Tab, New).
What Can I Say? Dialog Issues
What Can I Say? Dialog Opens Randomly
There is a known issue where background noise may cause the What Can I Say? dialog to open randomly. Likewise, background noise may cause some user commands, such as commands to browse webpages to be executed. To avoid this, turn the microphone off when you are not actively using SpeakToText.
Opening the “What Can I Say?” Dialog In Both SpeakToText and CoolInfo at the Same Time
If you open the “What Can I Say” dialog in both SpeakToText and CoolInfo, they are likely to open in the same place, so that one superimposes on the other. This makes it possible to confuse the “What Can I Say?” dialog of the two programs. If you are running both SpeakToText and CoolInfo together, it is best to close the What Can I Say dialog in one program before opening it in the other.
Opening the “What Can I Say?” Dialog From Advanced Mode
When opening the “What Can I Say” dialog in Advanced Mode, the “What Can I Say?” dialog opens at a reduced size. If it is left at this reduced size and not expanded with the mouse, it is more easily lost behind SpeakToText or other programs. Under this scenario, if you cannot find the “What Can I Say” dialog, try using the “Close Me” speech command, then reopen it with the “What Can I Say” speech command. If you still cannot locate the “What Can I Say” dialog, try closing and restarting SpeakToText.
Turning the Microphone Off While Speaking
If you turn the microphone off while you are speaking, the speech engine will stop recognition but the microphone button will not respond until you stop speaking. This occurs because the speech engine does not release the microphone while it is receiving input. To avoid this problem, stop speaking when you turn the microphone off.
Clearing the Speech Buffer While Speaking
If you press the Clear Speech Buffer button (or the “esc” key) to cancel speech recognition and clear the speech buffer, this stops the speech recognition engine turns the microphone off. Note that although this stops speech recognition, the Microphone will not turn off and SpeakToText will not respond until you stop speaking. This occurs because the speech engine does not release the microphone while it is receiving input. To avoid this problem, stop speaking when you clear the speech buffer.
The same thing happens if you are processing a .wav file and you press the Clear Speech Buffer button (or the “esc” key) to cancel the process, the microphone may not turn off until the speech engine encounters a pause or has completed cycling through the .wav file. This is related to the speech recognition engine not releasing the microphone while it is receiving input. Be patient, and the microphone will turn off and SpeakToText will become responsive again.
Numbers Are Not Always Formatted Correctly
Numbers between 1,000 and 9999 are not always formatted correctly. For example, six thousand eight hundred ninety one comes back as 6891 instead of the desired 6,891. This is a speech engine issue over which we have no control, and the recommended workaround is to manually place the comma with the mouse and keyboard.
Time of Day Not Always Formatted Correctly
When speaking the time using AM or PM, morning (AM) times are not always formatted correctly. For example, four sixteen am may come out as 416 am instead of the desired 4:16 AM. This is a speech engine issue over which we have no control, and the recommended workaround is to manually correct the time using the mouse and keyboard.
A Banner Still Appears After Entering the Registration Key
After you enter your registration key and continue, the second banner still appears. This occurs only once, and it occurs because SpeakToText was launched from the Trial startup dialog. The next time you launch SpeakToTextÔ 2.0, the Trial dialog will not appear, and SpeakToText will be banner free.
New Line or New Paragraph Command Sends Message
When using SpeakToTextÔ 2.0 with MSN Messenger, the new line or new paragraph command will send the message in MSN Messenger. This happens because the enter key in MSN Messenger sends your message. In SpeakToTextÔ 2.0, the new line command sends the enter character once, and the new paragraph command sends the enter key twice. This also applies to other chat applications that use the “enter” key to send a message. The solution is not to use the “new line” or “new paragraph” speech commands when dictating chat messages. Note that in MSN Messenger, if you want to start a new line, press the “ctrl” and “enter” keys on your keyboard simultaneously. Pressing these keys twice will start a new paragraph.
“Scratch Last” and “no no” Commands Do Not Always Work
The “Scratch Last” and “no no” commands work only with spoken text. They do not work on typed or reformatted text that has been i)typed from the keyboard, ii)pasted from the clipboard or iii)restored with the “Undo Scratch” command. The same applies to spoken text that has been re-formatted by SpeakToText™ 2.0. Reformatted text is any text that has been received from the SAPI speech engine, and changed by SpeakToText, for example numbers or text that has been capitalized by the “cap all text” command.
Another issue is where the “scratch last” and “no no” commands do not always work following a paragraph break. This behavior is inconsistent with normal text, and consistent with text dictated when “caps lock” is toggled on. The recommended workaround is to use “select sentence/paragraph” command followed by the “delete selection” command.
If “scratch Last” and “no no” are not working, try using the “select sentence” or “select paragraph” commands followed by the “delete selection” command to delete the desired text. Of course, you can always use your mouse and keyboard to select and delete text.
“Select Last” Command Does Not Always Work
This is similar to the issue above with “scratch last.” The “select last” command works only with spoken text. It does not work on text that has been typed from the keyboard or pasted from the clipboard. The same applies to spoken text that has been re-formatted by SpeakToText™ 2.0, for example reformatting of numbers or text that has been capitalized by the “cap all text” command.
If “select last” is not working, try using the “select sentence” or “select paragraph” commands. Of course, you can always use your mouse and keyboard to select text.
Clear Word and Scratch Word Speech Commands Do Not Always Work
There is a known issue with the speech commands “Clear Word” and “Scratch Word.” If either are used in SpeakToText after the cursor has been moved, either command will work once the next time it is invoked, but will not work again after that. If the cursor is again moved to another place in SpeakToText (other than the end of the document), then invoking either command again will delete the word to the right of the cursor (instead of the word to the left as it does under normal conditions). It is best not to use the “Scratch Word” or “Clear Word” commands if the cursor has been moved manually with the mouse, or by speech with the “go to top/bottom” or “go to beginning/end” speech commands.
Undo Last Scratch Does Not Always Work
In order for the "undo scratch" command to work, it must be used before you insert any additional new text. If you say something after scratching text, or if the microphone picks up background noise and recognizes it as a word, the undo scratch will no longer restore the deleted selection because new text has replaced it.
The cause and result of this problem can be best illustrated by a comparison to browsing a series of WebPages. If you start at page A, go to page B, then page C, you can move back and forth between pages A, B and C. But if you are on page B and you go to page D, you can no longer get back to page C using the back button because page C has been replaced by page D, and if you use the forward button from page B, it will take to page D because page C has been removed from the sequence.
Similarly, in SpeakToText, using words A, B and C as an example. If you scratch word C, then word B will be the last word. You will be able to undo the scratch of word C only until you speak word D. Once word D is spoken, you cannot go back and restore (undo the scratch) of word C because word D has now taken its place.
The “Click Focus” command doesn’t work after using the “Copy to Word” command
After you use the “Copy to Word” command or button, if Word was not open, the “click focus” command will not be active. In this case, you need to use the mouse and click on SpeakToText in the taskbar in order to bring SpeakToText back to focus. After this, the “click focus” command will function normally. This only happens when Word was not already open when you used the “Copy to Word” command or button.
The “Force Num” Speech Command doesn’t work
After saying the “force num” command, the command is not recognized, or the number (between zero and twenty) is still spelled out.
You MUST pause briefly after saying the “force num” command and before saying the number.
Clear Speech Buffer using the “Esc” key doesn’t work
In order for the “Esc” key to clear the speech buffer, SpeakToText MUST have the active focus. Otherwise, the application with the active focus will receive the command from the “Esc” key.
Before pressing the “Esc” key, click on SpeakToText in the taskbar, or click directly on SpeakToText.
Language Settings and SpeakToText
When using SpeakToText with the system language set to other than US ENGLISH, some unwanted character translation may occur. Using SpeakToText with other than US ENGLISH is not recommended.
Copying from the Clipboard
Pasting from the clipboard when it contains graphics and characters other than text has been known to result in failure of SpeakToText on some systems, and could cause you to lose your work. While this is not typical, we do not recommend pasting from the clipboard when it contains graphics and characters other than text.
The Microphone and Text to Speech
When you use any of the “Read” commands, SpeakToText turns the microphone off while it is reading your selection. This prevents the speech recognition engine from trying to recognize the words that are being read. When SpeakToText is finished reading, or when you use the “stop reading” button, the microphone is turned on again. If you use the Settings button while SpeakToText is reading, the microphone does not come back on when reading is finished, and you need to click on the Microphone button to turn it back on.
Application Selection Overwrite Protection (Advanced Mode)
SpeakToText now has a feature to prevent selected text in the target application from being overwritten. The default setting is enabled. Side effects of this feature are that text can only be inserted into the target application at the end of the text, and it causes some applications to beep when text is being dictated. This feature can be disabled using the “Insert” key on your keyboard, or from the General tab in the Settings dialog.
When using the “insert” key to disable/enable overwrite protection, SpeakToText MUST have the focus when you press the insert key. Otherwise, the “insert” key will act on the application that does have the active focus. To do this, click on SpeakToText (either on the SpeakToText window or in the taskbar) to bring SpeakToText to focus. Press the insert key, then go back to the target application in the same manner and place the cursor where you want your dictation to appear.
If you disable this overwrite protection feature, be sure to save your work before selecting text, or make sure either listening or the microphone in SpeakToText are turned off first. If selected text is unintentionally deleted, use the undo button in the target application to recover the text.
Running Two Copies of SpeakToText
When starting, SpeakToText checks to see if another copy is running, and if so it does not allow the second copy to start. This check does not work if a second copy of SpeakToText is launched while the first copy is still starting, because it is not fully initialized.
Also, in configurations where the default speech recognition profile is selected in Speech Properties, SpeakToText displays a dialog to prompt you to train your own speech recognition profile, and suspends the startup process until the user cancels out of the dialog, or enters and exits Speech Properties. Under these conditions, while startup is suspended it is possible to start a second copy of SpeakToText, although we do not recommend or support this. Running more than one copy of SpeakToText at one time may cause unusual behavior or a crash in one or both copies. This is not recommended and not supported.