About Us    Products    Store    Support    Download    Contact 

SpeechStudio RELEASE NOTES

SpeechStudio products were first deployed in November 2000.  Here are some highlights of their progress, by (selected) version.

 

Release Highlights for Version 3.8.1 (December 31, 2005)

General

  • Version 3.8.1 is a bug fix release of SpeechStudio Suite v3.8.0. It is upward compatible with v 3.5.* thru 3.8.0.
  • Samples are now supplied, as part of the <language>\Events sample, to show how to program in Dragon Dictation in cooperation with SpeechPlayer.

Incompatibilities with Version 3.8.0

  • You are advised to rewrite your Dragon control code to work through the new SpeechPlayer Dragon properties listed below for the SpeechStudio Voice Control.
  • Message2U event name Message2U_DragonMutingChanged has been changed to Message2U_DragonStatusChanged, and now covers a variety of changes in the status of Dragon, not just its muting.  This change reflects the broadened capability of SpeechPlayer to control Dragon Dictation.

SpeechPlayer

  • [Fix] In some cases, SpeechPlayer would not LocateControl's properly if they were within a C# or VB.NET dialog group box; that is repaired in 3.8.1.
  • You can now tell what status wav is used by going to SpeechPlayer's Tools | Options and clicking into the "Feedback Options".  A "*" will be in front of the current state.
  • When directed to establish a "dragon target window" (via GetPropertyNum), SpeechPlayer will try to move Windows focus into that target window and will bring its parent Window to the top.  This is because Dragon only dictates to the topmost window.

SpeechStudio Voice Control

  • [Fix] The SpeechStudio Voice Control now allows control of ScanSoft/Nuance Dragon through SpeechPlayer, via property values.  This allows Dragon to be controlled through a single API instance, instead of having one API instance in SpeechPlayer and then others in various user applications.  Dragon was unstable when controlled from more than one API instance, and would often hang.  This approach also allows development of basic Dragon dictation support without programmer involvement with the Dragon API..
  • Set.../GetPropertyNum now supports "dragon connected", so you can programmatically begin connecting to Dragon in advance of requesting dictation; Dragon is slow to initialize, and so this creates better user experience.  It also controls Dragon disconnect.
  • Set.../GetPropertyNum now supports "dragon target window", so you can programmatically direct Dragon dictation to any window that supports text capabilities.
  • Set.../GetPropertyNum now supports "dragon microphone on", so you can programmatically control Dragon listening.  Note that Dragon can be slow to respond, so that turning off the microphone may not take effect for many seconds.

AudioTeeControl

  • [fix] If SpeechPlayer's Tools | Options | Sound Input "Allow standaside recording" was checked, then some explicit or implicit engine restarts could result in a very large initial threshold value, that caused blocking of further input.  The symptom was that the SpeechPlayer sound level histogram would say "Listening" but no histogram values.  The workaround is to turn off "Allow standaside...", exit options to force a restart, and then turn "allow standaside..." back on.  This bug is fixed in v3.8.1.

 

Release Highlights for Version 3.8.0 (December 8, 2005)

General

  • Version 3.8.0 is a major new release of SpeechStudio Suite. It is upward compatible with v 3.5.* thru 3.7.*. It introduces support for remote web clients and Dragon interoperation.

Incompatibilities with Versions Before 3.8.0

  • The SpeechUtilities.cs in the Samples/CSharp area has changed its namespace name from "SpeechUtilities" to "SpeechStudio".  Public class Speech has changed names to public class SpeechUtilities

SpeechPlayer

  • SpeechPlayer now supports parallel use of Nuance/ScanSoft's Dragon for Dictation with Microsoft's SAPI 5/6 engine for Command & Control.  Dragon Dictation is excellent but its command and control is rather rigid and slow.  Microsoft's command and control is fast and flexible, but it is weak for dictation.  The combination is strong for both.  SpeechPlayer Tools | Options allow end users to set up their own hotkeys to switch between engines.  New Message2U events to alert your applications of changes in Microphone control to/from Dragon, allowing them to set focus for dictation, for example.  A Message2U event is now provided for changes in SpeechPlayer muting as well.
  • SpeechPlayer now provides a "Reconnect to Dragon" button in its Advanced Settings under the "Sleeping" tab of Tools | Options.  This button allows you to shut down Dragon while your SpeechStudio apps are running, and then later reconnect to Dragon if you bring it back up.
  • SpeechPlayer uses multi-threading to protect your Apps from the sometimes long delays inherent in Dragon's dictation, even when natspeak.exe hangs.
  • SpeechPlayer now can generate a background audio reminder periodically to let the user it is listening for something.  The prompt sounds are configurable by the end user, as their frequency.  Separate sounds can be issued depending on the state of SpeechPlayer, such as "listening for commands" or "listening for commands and Dragon is listening for dictation at the same time".
  • SpeechPlayer will now look for a focus-less client (one started with an Init call with a 0 hWnd) if not focus-driven process should have speech focus.  This allows background grammars to persist and recover after focus-driven apps give up control.
  • Allow the user to set up a remote microphone from SpeechPlayer.
  • Some audio devices will buffer up substantial amounts of sound, so that when muting ends, several seconds of buffered sound could be sent to the engine. SpeechPlayer now cleans out buffered sound so that actual listening starts about the same time as muting is clicked off.
  • If SpeechPlayer was dealing with an engine (say one named "English Recognizer") and that engine was de-installed, then SpeechPlayer reported that it could not find a SAPI engine when in fact there may be other engines available. It now explains that its old engine is not available, and asks the user to choose a new engine.
  • SpeechPlayer can now send you a "Message2U" event with a code of Message2U_FalseRecognition indicating a false recognition. Its "text" field has a guess at what was heard followed after "??" with a string explaining why the recognition failed:
    "NOISE", "MIKE", "LOUD", "SOFT", "FAST", "SLOW"
    You'll need to set the "Deliver False Recognitions" property to 1 to receive these messages.
  • SpeechPlayer can now generate TTS to a file, with its SpeakToFile command.
     

SpeechStudio Voice Control

  • ListenForSPProxy command allows licensed user to tell SpeechPlayer to listen for a remote proxy.
  • UseRemoteMic command allows licensed user to tell SpeechPlayer to connect to a remote microphone object - as supplied by SpeechStudio.  SpeechPlayer will then get its sound from that remote device.
  • For the Speak() method, when using SPEAKFLAGS the first character after the closing ']' was ignored. For example, "[SPEAKFLAGS:SPF_IS_XML]what's up" would be spoken as "hat's up". The workaround is to put in a space, for example, "[SPEAKFLAGS:SPF_IS_XML] what's up".
  • Set.../GetPropertyNum now supports "enable sound feedback", so you can programmatically disable sound feedback.
  • Set.../GetPropertyNum now supports "deliver false recognitions", so you can programmatically get information about poor recognitions delivered to your App.
     

SpeechStudio Remote Microphone

  • The Remote Microphone allows the speech engine on one machine to do speech recognition on sound from a remote machine.  The Remote Microphone can run on any Windows or Windows CE device.
  • The remote microphone can be told a range of ports, allowing it to support firewall tunneling.

SpeechStudio Proxy

  • The Proxy allows SpeechStudio clients on one machine to control a SpeechPlayer on a remote machine.  Typical applications include web programs and terminal server applications.
  • Tracks window activation, destruction, focus and keypress actions on the client side, passing them on to the remote SpeechPlayer so it can track focus. That way grammars on the SpeechPlayer will automatically stay in sync with changes in the remote application..
  • The Proxy can be told a range of ports, allowing it to support firewall tunneling.
     

SpeechStudio

  • SpeechStudio can now handle more structural changes to VB projects without complaints.

SpeechRunner

  • You can now specify arguments for the programs that you launch from SpeechRunner.  An argument edit box in included in the Setup Dialog and in the SQR Command dialog.

UI (User Interface) Control

  • SpeechStudio's UI Control provides the WindowCreationParameters command, allowing you to specify some important attributes of predefined SAPI 5 dialogs and wizards before they are launched. These dialogs can now be launched asynchronously, so that your code can maintain control while the dialog is active. You can specify their initial position and to some extent their names. And you can control their behavior such as, for example, having them be always on top.
  • You can now use the StandardUserTraining command to launch user training with the engine vendor's training texts. You can use the UserTraining command to launch training with your own specific texts.

AudioEditControl

  • [fix] Truncation sometimes happened on odd bytes, if the user specified it.  Any appended sound then had upper and lower bytes reversed - generating noise.  We now prevent odd byte splitting on truncation.

AudioFilesControl

  • The value passed to SaveAs is not longer required to be VARIANT_TRUE or VARIANT_FALSE. Now simple true or false is accepted. Previously, using "true" could sometimes fail to delete temp wav files.

Grammars

  • The SpeechStudio.2Digits grammars now include In_20_to_29, In_30_to_39 and so on to In_90_to_99, so that users can compose their own grammars more easily.
  • The SpeechStudio.DateTime grammars now include time_to_the_minute (hours, and optionally 0, 1, 2, ...59 minutes), and minute_phrase (0, 1, 2, ...59 minutes) in addition to the approx_minutes versions.

 

 

Release Highlights for Version 3.7.3 (November 18, 2004)

Version 3.7.3 is a support release of SpeechStudio contract work and future development.  It has minor bug fixes and improvements for general users.

Release Highlights for Version 3.7.2 (February 15, 2005)

Version 3.7.2 is a support release of SpeechStudio contract work and future development.  It has minor bug fixes and improvements for general users.

 

Release Highlights for Version 3.7.1 (January 6, 2004)

SpeechPlayer

  • (Bug Fix) In some cases, dictation with non-base logging levels could cause a fault in SpeechPlayer.
  • (Bug Fix) Dictation focus would not always track with changes in edit window focus for multiple dictation windows in a single client.
  • Logging output was reduced and refined..
  • SpeechPlayer now issues Message2U callbacks to let the app know when muting has changed.  Users can now create their own notifications of mute status. such as greying their application.

SpeechStudio Voice Control

  • SetPropertyString("announce confirmations", <0 or 1>) is now supported; you can programmatically have SpeechPlayer request confirmations by saying "please confirm <command>".

 

 

Release Highlights for Version 3.7.0 (December 10, 2004)

SpeechPlayer

  • (Bug Fix)  Microsoft XP Upgrade SP 2 has a bug in its get_hFont routine; it showed up as a crash in SpeechPlayer, typically in the Microsoft's oleaut32.dll module.  The workaround was to turn off "Show Misrecognitions..." on the Recognitions tab of SpeechPlayer's Tools | Options.  The bug was triggered when a misrecognition report tried to change font.  Version 3.7 avoids all calls to get_hFont.
  • (Fix)  If the user selected Tools | Options twice, SpeechPlayer asserted. Now it will just reshow the existing options box.
  • SpeechPlayer no longer stops because an internal parameter is out of range; it can correct and continue with just a log message.

SpeechStudio Voice Control

  • (Bug Fix) The Control Log Name property could cause an infinite loop crash: if two program instances set the control log to the same name, then the second opener might loop trying to report the error by opening the log file - thus causing the error again.  The workaround was to avoid using the same log name in two programs.  Version 3.7 suppresses logging for the second program.

  • (Fix)  For the SetPropertyNum function, Auto Sleep should have been Auto Sleep Seconds.  Auto Shutdown should have been Auto Sleep Shutdown.  Neither of these properties worked correctly in v3.6.4.

SpeechRunner

  • SpeechRunner can now launch its test programs with arguments.  In previous versions, a command file was used as an intermediary to provide arguments.  This had the drawback that the program under test was the command file invocation rather than the program itself.  In Version 3.7, the Default Settings command and the Settings dialog allow an argument string.  The target program can then be launched directly.

Release Highlights for Version 3.6.6 (October 1, 2004)

Incompatibilities with Versions before 3.6.6

  • The WatchForWindow command to the SpeechStudioControl now has an additional required parameter: bEverywhere.  If this parameter is true, SpeechPlayer will report windows matching the given pattern regardless of what process they arise in.  If it is false, only windows in the same process are considered - this was the only choice in versions before 3.6.6.

SpeechPlayer

  • (Bug Fix)  The Transparency Pattern was being applied to windows even when they have their own client.  This bug showed up when clicking from a non-client "transparent" window back into the current current client window:  SpeechPlayer might stop listening.
  • SpeechPlayer's Tools | Options now support a Sound Input tab page, on which you can set up connection to a SpeechStudio Remote Microphone or other instance of the SpeechStudio AudioSendControl.  When selected, sound will come from that remote microphone.
  • (Fix) A window can be activated many times.  SpeechPlayer incorrectly sent a WindowDestroyed message for each activation; it now sends just one WindowDestroyed message.
     

SpeechRunner

  • SpeechRunner can now simulate the basic mouse commands.  The command "DoMouseOp: Click, (24,543)" will cause the mouse to reposition to x=24 and y=543 relative to the rectangle for the current SetTestFocus window; the current SetTestFocus window will be brought to the top (if possible) and the events (down and then up) of a "click of the left mouse button" will be simulated.  Move, DoubleClick, Down, Up, RightClick, RightDoubleClick, RightDown and RightUp are also supported.  These commands may be used to test graphics or to push old-style buttons that do not have window substructure.
  • The SetTestFocus command now allows the selection of an unnamed window.  The window name pattern can be used to select some named relative of the desired window; then a prefix of the form [child.sibling.sibling.child] can walk to the desired unnamed window.  Besides sibling and child, owner and parent relationships are available.
  • If the user starts another test while one is still running, SpeechRunner now disconnects the older test, recovers, and continues with the new test.
  • If the user tries to start a test which does not exist or is not usable, SpeechRunner now complains, recovers and continues; previously, SpeechRunner just waited for the test to become usable.
  • (Fix) In rare cases test disconnect happened while cleanup was incomplete, causing the end of test message to be suppressed.
  • (Fix) The DoMenuClick, DoMouseOp, AwaitWindow, AwaitWindowGone, AwaitLogText, and
    AwaitSpeak commands no longer require that if an engine is present that it be listening.  These commands now operate independently of the engine state.
     

SpeechStudio Voice Control

  • The Control Log Name property can be set to a simple root file name in order to capture logging in your application’s controls.  All logging messages for control in your app will go to a file named MyControlLog.txt located with the other SpeechStudio logs, in a directory defined in SpeechPlayer’s help.  Give only the file name root; the extension is always .txt and the directory is always the SpeechStudio log directory.  A file name of “” will turn off logging.  Each log line will be prefixed by the number of the control that issued the log message.

  • The WatchForWindow command can look at all windows, not just those in your own process.  The bEverywhere parameter controls this behavior.

  • SetPropertyString("engine", "new recognizer name") is now supported; the special engine name "SuppressEngine" allows the user to disconnect SpeechPlayer from any engine. This allows the user to compensate for engine bugs where the engine crashes if it is left running too long.

 

 

Release Highlights for Version 3.6.5 (April 27, 2004)

General

  • Version 3.6.5 is a upgrade for v3.6.4, and offers improved support for Audio Recording. It is upward compatible with v 3.5.*, 3.6.1, 3.6.2 and 3.6.3.
  • Audio recording can now be continued independently of the engine listening status.
  • Recording can now elide sub-threshold sound.
  • Introduced the Audio Edit Control: delete sound data after a given point to accomplish "rewind" functionality. See the Recorder sample under Samples/Cpp or Samples/VB for usage.

Profile Developer

  • Profile Developer users will now be able to control licensing on remote machines. Ask SpeechStudio Support for details.

SpeechStudio Voice Control

  • The "DisplayMode", "Auto Sleep Seconds" and "Auto Sleep Shutdown" properties did not work correctly in v3.6.4.

 

Release Highlights for Version 3.6.4 (November 25, 2003)

General

  • Version 3.6.4 is a upgrade for v3.6.3, and offers tuned support for some popular telephony devices. It is upward compatible with v 3.5.*, 3.6.1, 3.6.2 and 3.6.3.
  • Some suggestions are now available for those who are using or designing for PC Telephony, for example with the Zoom 3025 modem family or Intel Dialogic cards.  See Start | Programs | SpeechStudio | Help | Hooking Up Phones.
  • Added a modem test program for end users, to help them qualify and set up a modem.  See Start | Programs | SpeechStudio | Tools | Test Modem.
  • A bug was reported in the order of speechstudio3.h declarations. The SetRemoter (unreleased) operation was out of order. The effect is that WatchForWindow, LocateControl, ListenToFile, and DictationActivatedFor methods would sometimes fail.

SpeechPlayer

  • SpeechPlayer now allows the end user to adjust rate and volume of TTS.  Pitch is controllable only in SAPI 4.  A TTS pause hotkey is displayed but not implemented.
  • SpeechPlayer's TTS support for SAPI 4 has been improved to allow display and selection of voices as well as control of speed and volume.
  • SpeechPlayer now allows WhatCanYouSay to be null.
  • SpeechPlayer no longer goes into sleep mode if the user has muted SpeechPlayer.
  • SpeechPlayer will no longer sleep in cases where we are hiding a speech client with a non-speech-enabled window from the same process.  Now SpeechPlayer will stop listening entirely until a voice client regains control.  Exception:  when the telephone is used for input and the developer has set the property "GiveFocus to RequestLine app" then control will immediately switch to the RequestLine master app.
  • SpeechPlayer and its engine management have been accelerated for large grammars. 
  • You now may allow a non-speech window to get control while listening for another (speech-enabled) window:  SetPropertyString("transparency pattern", pattern).
  • Message2U codes now reserve 1..100 for SpeechPlayer messages for odd synchronizations, error and warning reports.  You can tell quickly about error-type messages.
  • SpeechPlayer now issues Message2U callbacks to let the app know when the ListenToFile call has completed.  In previous versions, the user could only guess at this, and if a mistake was made it could hang the engine.
  • SpeechPlayer now pads wav file inputs (SAPI 5 only) with silence at their front and back to give the engine time to see a pause at the beginning and end. This adjustment improved recognitions from directly recorded files by about 30%, to bring it up to about the levels seen from the initial live input.
  • The SimPhone panel (to simulate phone calls) now displays when the user selects SimPhone option - and is removed when it is unselected. Now users can try SimPhone without having to program a special trigger into their app.
  • If a client's dynamic grammars can accept a phrase ambiguously, SpeechPlayer now dumps out each parse so the user can see where the ambiguity arises.

Dictation Manager

  • Dictation Manager will now accept a "point" in its number editor, allowing decimal numbers.
  • Dictation Manager will now restore focus correctly after spelling or number entry, and will maintain focus during those operations so that the cursor and selection is visible.
  • Dictation Manager has improved its text formatting.

Profile Manager

  • Profile Manager is now included in SpeechStudio Suite.  It will have rights based on its Free Software license, unless the full license is purchased separately.

SpeechStudio Voice Control

  • The EnableDictation method can now create managed or unmanaged dictation that is disconnected from Windows focus.  You control its enable/disable directly.  The user can be dictating while doing other actions on Windows.
  • You can now control of rate and volume of TTS through properties "TTS rate" and "TTS volume", programmatically.  A "TTS pitch" parameter is effective only for SAPI 4 engines.
  • You can now pause and resume TTS through the property "TTS paused".
  • You can now get the list of available SAPI 4 voices in the "voice list" property.
  • You can now control when grammars are refreshed, using the "Do not refresh" property.  When this property is set true, then incidental refreshes, such as those normally generated on focus switches, will be suppressed for that specific client.  This can improve performance of very large grammars.

 

Incompatibilities with Versions before 3.6.4

  • Windows that are part of the same process are now considered to hide the current speech-client even if they are not direct descendents.  If you want a client to keep listening when a hiding window (of this process) gets focus, then use the "transparency pattern" parameter to describe the hiding window as transparent.

SpeechRunner

  • SpeechRunner can now simulate the dictation commands for text formating and punctuation.  The command "Say: new paragraph", for example, now has the expected semantics, following the effects of typical dictation engines.

 

 

Release Highlights for Version 3.6.3 (July 22, 2003)

General

  • Version 3.6.3 is a bug fix release for v3.6.2 and a tuning release for the Voice Tracker array microphone (see www.acousticmagic.com). It is upward compatible with v 3.5.*, 3.6.1 and 3.6.2.
  • A Tutorial is now available for those who are using or designing for the Voice Tracker array microphone. See Start | Product Files | SpeechStudio | Tutorials | Using Voice Tracker.
  • The threshold calculations were found to be in error - the effect is that noise was not suppressed unless the threshold was set very high.
  • This version of SpeechPlayer supports user "sessions". The SAPI 5 engines do not appear to resume correctly after switching users (e.g. on XP). In this version, SpeechPlayer shuts down the underlying engine for one user and restarts it after a new user session is selected, as needed. The transition should appear seemless, except that on restart the engine will be muted.
  • Version 3.6.3 includes some additional support for SAPI 4 engines such as Dragon/ScanSoft Naturally Speaking. If you try a SAPI4 engine, please do not use the thresholding Audio object - a warning will remind you. The audio object formating is not compatible with at least the Dragon engine and hurts accuracy.

 

Release Highlights for Version 3.6.2 (June 20, 2003)

General

  • Version 3.6.2 is a bug fix release for v3.6.1. It is upward compatible with v 3.5.* and 3.6.1.
  • No bugs were found in the v3.6.1 beta test. Some documentation errors have been fixed and some clarifications added.
  • Version 3.6.2 includes some internal tuning in support of the AcousticMagic Voice Tracker(TM) array microphone.

 

Release Highlights for Version 3.6.1 (June 7, 2003)

General

  • Version 3.6 is primarily an end user upgrade for v3.5*. It is upward compatible with v 3.5.0, 3.5.1 and 3.5.2.
  • Version 3.6 is the first SpeechStudio product to find and allow limited use of SAPI 4.1 engines as well as SAPI 5 engines. SAPI 4.1 performance and correctness is not warranted. You may switch between engines by using SpeechPlayer's Tools | Options on the Recognition tab, but you'll need to select that same tab again to set up the profile, since the profile combo box does not reset for the new engine until that engine is actually selected - i.e., you've close that page with "ok". SAPI 4 support is limited: it does not support dictation, for example.

Incompatibilities with Versions before 3.6.1

  • Users of Dictation Manager will find its behavior changed slightly. Most edit commands are now available only in edit mode. In edit mode, dictation is suspended. This change was made to remedy user complaints of degraded dictation accuracy.

SpeechStudio Voice Control

  • The SpeechStudio Control can now compile grammars on the fly for either SAPI 5 XML format or SAPI 4 cfg format. This retargeting should be invisible to you and to your end user.
  • Added support for the WindowDestroyed event. This even is generated once for any window that is previously reported by a NewWindowActivating event, at the time that window dies. The NewWindowActivating event is restricted by the WatchForWindow pattern, but WindowDestroyed is independent of later changes to the window name. Together these facilities allow you to get voice focus back and clean up after a foreign window is closed.
  • Added support for the DictationActivatedFor call. This call can be used to suspend dictation for a client, while still receiving the managed window focus events.

SpeechPlayer

  • SpeechPlayer now supports Press-to-talk. Select press-to-talk behavior for your hot key by going to Tools | Options and looking on the Sleeping tab for the Press-to-talk radio button. Note that the hot key will reverse the current mute state of SpeechPlayer, so if SpeechPlayer is Listening, holding the hot key will MUTE.
  • SpeechPlayer now supports automatic sleeping. Sleeping means that SpeechPlayer will stop all application grammars and listen only for a single "wake up command". You can control if, when and how this happens. Just go to the Sleeping tab under Tools | Options. Sleeping is different from muting, in that when SpeechPlayer is sleeping it is still listening, although just for a single command; that command will not do anything drastic, it just restores the previous listeners. The idea of sleep is to keep SpeechPlayer from recognizing some dangerous command by accident, while still allowing voice control. Muting, on the other hand, turns off voice; muting cannot be ended by voice, since voice is turned off.
  • SpeechPlayer now supports automatic muting. For example, if you want SpeechPlayer to mute itself after 30 seconds of inactivity, then go to the Sleeping tab under Tools | Options; check to box for "Automatically sleep (or mute)..." and put "30" for seconds. The clear the "wake up command" edit box.
  • Bug fix: Restore the registry's value for the voice threshold when starting a new engine.
  • Version 3.6 reorganized the entire threshholding process. The problem was basically that the user could drop the mike or otherwise generate a loud noise which then would cause the max possible volume to be set very high - that caused the threshold to go too high, so everything was suppressed thereafter. The new version has SpeechPlayer call in whenever there is a successful recognition; the audio object keeps track of the last few seconds of input and calculates a new max from that history, on the assumption that if the engine could recognize it, then it wasn't a hyper-loud noise such as a mike drop.
  • The gotosleep and wakeup specifications to user-accessible properties so they can be set programmatically through new properties.
  • Add properties to tell/set when we are sleeping and when we are muting.
  • Bug fix: Do not SetPrompts for a client that is not listening.
  • Do not overwhelm the log with useless "SAPIEVENT" lines, since there are many.
  • Added "auto shutdown seconds" controls the time SpeechPlayer waits before it will shut itself down; any new client will inhibit the shutdown. Autoshutdown allows SpeechStudio programs to be distributed to end-users without having a "sticky" instance of the engine and SpeechPlayer retained on their desktop.
  • Allow the user to see false recognition information, if desired. This info can help train the user to speak in an even cadence, not too loud or too soft, as well as explain what might be happening.
  • Added support to have the profiles listed and restored directly from SpeechPlayer (Profile Manager must still be purchased separately.) The Profile file and auto-restore to are connected to registry options and so persist.
  • Added an option to allow SpeechPlayer to turn off screen and power saving modes. The .NET version will crash hard in some cases where SpeechPlayer is listening, gets a recognition, and tries to work with a control such as a button on a program that is asleep.
  • Bug fix: On some occasions a client would "disappear" immediately after its creation, if it immediately followed the death of another speech-client window. This was because of a misuse of the same memory from one client to the next. This is now prevented.
  • Bug fix: In cases where the user did not have auto-Answer selected and had no program listening to answer the phone or elected not to answer the phone, an incoming call was ignored until it disconnected; the disconnection caused the engine to switch to FileInput mode. This was wrong in any case, since any SpeechRunner test would have been stopped by the call if it was answered.
  • The version 3.6 log is cleaner, in that it does not print out grammars that are not active.
  • Tell the user immediately if the microphone can't be connected.
  • SpeechStudio can now suspend dictation while still maintaining focus control of its managed windows. The app developer can use this suspended mode to perform window-related grammar or mouse activities without complications from dictation recognitions.

Dictation Manager

  • Dictation Manager will now support separate modes for edit and dictation. Users reported that dictation accuracy was substantially reduced when complex grammars are presented along with the dictation engine, so in the new version dictation mode has only a very limited command grammar: correct that (enter correction mode), scratch that (delete the last chunk), start number mode, start spell mode, start edit mode, stop dictation ?mode.
  • Dictation mode now supports a "stop dictation" command. Also, dictation is turned off during edit mode.
  • Dictation Manager now uses What Can You Say to support organized help during the edit, spell and number modes. These grammars and usages were complex for end users.

Profile Manager

  • Profile Manager will now share information with SpeechPlayer directly. Starting in version 3.6.1, you may launch Profile Manager from SpeechPlayer's Tools | Options under the Profiles tab. Profile Manager will automatically hand information back to SpeechPlayer about your available profile backup choices.

SpeechRunner

  • Bug fix: On some occasions SpeechRunner tests would cause SpeechPlayer to produce a complaint box saying "Sorry, your phrase is too complex for SpeechPlayer to analyze" even though the phrase might be trivial. This was because of uninitialized memory. This is now repaired.

 

 

Release Notes for Version 3.5.2 (March 19, 2003)

General

  • Version 3.5.2 is primarily a bug fix release for version 3.5.0, and is compatible with v 3.5.0 and 3.5.1.
  • The biggest remaining issue area is with the .NET version;  Microsoft changed the message delivery to basic windows controls in .NET: basic controls like buttons will not receive messages such as BN_CLICK unless the control is "mouse visible".  As a result, SpeechStudio Control will try to make its target/client window mouse-visible before sending on any windows control messages.  You will see this as your speech-enabled windows popping up.

SpeechStudio Voice Control

  • SetForegroundWindow is now called prior to attempting control message delivery; in .NET, controls such as buttons are not given their messages unless they are in the foreground - i.e., visible for a potential mouse click. They do not have to be the active window.
  • Add new control method:
          ListenToFile( fileName )
    Causes the engine to be restarted (if necessary) in unshared mode with a file audio input, and loads the given file. Any activity, such as recognitions, will happen thereafter as though the microphone had been in use.  Upon completion, or if fileName  is "", the engine is restarted listening to the microphone.
  • Recent additions to send LogText messages from the control - designed to inform the user about control issues - could potentially hang. The problem was that the rare messages could be requested during a COM interaction with SpeechPlayer.

SpeechPlayer

  • SpeechPlayer will now reset the SPI_GETFOREGROUNDLOCKTIMEOUT system parameter to zero to shutdown the lock timeout.  This timeout normally prevents window switches when certain programs, such as notepad, lock the keyboard to their input. SpeechPlayer restores the old timeout when it is shut down.  This change allows voice action messages to reach "hidden" controls.
  • Add an option to allow SpeechPlayer to turn off screen and power saving modes. The .NET Speech engine will crash hard (require unplugging!) in some cases where SpeechPlayer is listening, gets a recognition, and tries to work with a control such as a button on a program that is asleep.  This option is "on" by default, so the screen saver will not be enabled if SpeechPlayer is listening for an active client.
  • The Tools | Options | Recognitions voice threshold parameter was not saved.  Now we update the voice threshold parameter into the registry.
  • Two new read-only properties are supplied:
          GetPropertyString("loaded grammars", strProperty)
          GetPropertyString("active grammars", strProperty)
    Each returns a string value: a newline-separated list of the grammars.
  • In rare cases an OnSpeaking(audio stop) event could try to send a message via a dead client, causing a crash.
  • If you destroy a control on your speech-enabled frame then in some cases the client of that frame would Stop Listening. (Bug introduced in v3.5.0,3..5.1; fixed in v3.5.2)
  • If the user changed a speaker or other engine attribute while SpeechPlayer was muted, the engine reset would start it listening again even though SpeechPlayer's mute still showed as on.
  • A crash sometimes occurred when a window doing active dictation correction was closed. This problem and any related problem of delayed message action have been fixed.
  • Try to explain to the user when the SAPI engine has hung in the most common case of it being unable to activate or deactivate a grammar.

SpeechStudio Grammar Editor

  • A menu item that has a name with a number in it, such as "menu item 1", caused a crash in SpeechStudio in some cases when Insert Pattern is used. The workaround is to use the control number of the item and create the pattern by hand. (v3.5.0,3..5.1; fixed in v3.5.2)

SpeechRunner

  • If a problem occurs during its diff operation, SpeechRunner will now dump out the highlights and the timestamp-stripped data that it was working on, so you can inspect it by hand to correct the problem.

 

Release Notes for Version 3.5.0 (February 14, 2003)

General

  • Version 3.5 has added support for Visual Studio .NET:   Visual Basic VB.NET, C++ and C#.
  • Version 3.5 has added support for Foreign Windows.  A foreign window is one for which you have no code or for which you have chosen to make no direct or derivation-based VUI changes.  You can now add voice control to manipulate any components that you call or that are called indirectly.
  • The biggest source of issues in Version 3.0.0 and 3.0.1 was in attempts to use non-compliant TAPI modems.  We could not fix this but we have made the error results more clear and have provided greater instruction on avoiding the problems.

Incompatibilities with Versions before 3.5.0

  • The Refresh event has been renamed Refreshing.  This change was required for support of  .NET, since the old name collided with the Refresh method.  The Refresh method remains as unchanged.  Impact:  you must find all uses of the name Refresh and, if they apply to the event, change them to Refreshing.  Also changing the data structure name to REFRESHING_EVENT_TYPE and the message number to DISPID_Refreshing. This is the change that's refreshing!
  • The semantics of RequestLine have changed. Now, the boolean return value indicates whether or not SpeechPlayer found a telephony device. A "true" return value does not mean that RequestLine succeeded.  A new TelephonyEvent: TeleponeLine is delivered when the RequestLine succeeds or fails.
  • For TreeView, we now provide regularized navigation:  {Up | Down} {Out | Over | Into}.  The DownInto command now will open the tree to the next level if it can.

SpeechStudio Grammar Editor

  • The LocateControl(descriptor, class) prefix function has been added to allow dynamic search for a control on a given frame.  The descriptor may name the target control directly, may name a neighbor control and a direction to the target control, or may give an X,Y position for the target control.  The class gives the general control class, such as "Button" or "TreeView".   For example, the action LocateControl("OK", Button).Press() would press the OK button.
  • LocateControl(...) may be used for actions or for variables, for dynamic grammar creation.
  • For LocateControl(descriptor, class) the descriptor may be a pattern variable, and so may map to a control name at runtime.
  • Added a special dialog to help users build their LocateControl calls.
  • Fixed bug where the Grammar Editor sometimes crashed if opened without resources.
  • You can now double-click on a grammar (.grm) file to launch SpeechStudio.  It will locate the co-resident sqz file automatically.

SpeechStudio Voice Control

  • Don't refresh the grammar if the window is dead. This can happen now with greater frequency because of ForeignWindows, which can disappear under the grammar. Note: refresh doesn't crash, it just puts out warnings about any LocateControls that failed.
  • MapInsert(map, "", "written form") will now be construed to mean that the spoken form is the same as the written form.
  • Implement LocateControl as a method of our SpeechStduioControl. This method parallels LocateMethod in the grammar editor, and allows our user to access foreign window controls using the same syntax as for SpeechRunner and SpeechStudio actions.
  • Add WatchForWindow method to watch for the activation of a window whose name matches a given pattern.  Add the NewWindowActivating  event so that the user app can be signaled when a matching widow is activated.
  • Add GetWhatCanYouSay() to fetch prompts, grammar samples programmatically.
  • Add SetWhatCanYouSay() to control SpeechPlayer's prompts in its WhatCanYouSay window.
  • Add Message2U event to allow SpeechPlayer to provide error and informational messages asynchronously.
  • Tell the user what dat file was used in a SpeechPlayer log message.
  • Define a flexible lookup to find the .dat file.  Use the given directory/exe name as a path to find the dat file. If the exe is in /a/b/c/bin/foo.exe and we are told to find "foo.dat", then look first in /a/b/c/bin/foo.dat, then for /a/b/c/foo.dat, then for /a/b/foo.dat and so on. This allows C++ to be in any of the subdirectories (such as Debug or Release). More importantly, it allows C# users to not worry that their exe goes into ../bin/Debug or wherever. Users that care can give a full path name.
  • If no suffix is given for the dat file name, then add .dat.
  • Changed the Speak method to allow users to set up SPEAKFLAGS in SAPI.  Users can now pass through XML commands to control pitch, volume and so forth, in TTS.

SpeechPlayer

  • ResetEngine() has been no actually resetting in most cases. It was "optimized" to avoid reset if it judged that the engine was already in the desired mode. By forcing it to restart the engine, users can change engine options, such as the use of the audio tee object, on the fly.
  • Fixed PRN 301:  If the user calls Dial without calling RequestLine, we crashed.
  • We now do more to explain any grammar error to our user.
  • SimPhone variable is now a parameter, accessible via GetParameterNum.
  • Only deactivate a dictation grammar if it was started. Otherwise the log will be sprinkled with misleading deactivate messages.
  • The user can see what the current state of the grammar's prompts is. GetWhatCanYouSay(grammar_name).
  • The user can control a grammar's prompt display: SetWhatCanYouSay(grammar_name, string_of_new_prompts)
  • Added the QueryPrompt event: it is fired when the end user double clicks on a prompt in SpeechPlayer's WhatCanYouSay window.  You can handle this event to explain or expand on prompts, thus providing dynamic control of the SpeechPlayer WhatCanYouSay prompt display.
  • Changed the handling of HalfDuplex so that we stop listening but continue to process the message loop. We restore the system to listening when the sound output is over (as indicated by the AudioStop event.)  Speak and SpeakWav will now queue so the engine will not have to switch back and forth from listening to not listening, which in previous implementations cause a noticeable pause (.5-2 sec) between Speak calls.
  • DTMF can now interrupt (barge-in) during speak and speakWav calls, even in HalfDuplex mode.
  • The Microsoft engine will occasionally die/hang after responding "BUSY". We will now repeatedly try the rejected operation to see if it will get unbusy.  The problem is usually that the engine refuses to listen.  This happens when the profile is corrupted or on some occasions where the number of active grammars goes to zero temporarily: Occasionally when the number of active grammars goes to zero (for example when we are switching grammars,) SAPI may change the state to SPRST_INACTIVE. Thereafter SAPI will keep changing the state back to SPRST_INACTIVE no matter what we do, effectively hanging the engine. To work around this problem, we reset the engine under these circumstances, then restore the dynamic state.
  • We no longer complain about LocateSibling unless we actually need to use that result. Otherwise it confused the log.
  • If the program answers a call, we must unmute. otherwise the phone caller will get dead air. We then restore the state of muting after the call is over.
  • Add an option so that the AudioTeeObject is by default not used, and can be used by setting an option.
  • Implemented Message2U event so that errors/conditions can be reported back to the user program. It reports either to the control program if we are managing a telephone, or to the current client otherwise.
  • Changed many error/assert conditions to instead report their error as Message2U events, reducing MessageBox annoyance and assertion checks.
  • SpeechPlayer no longer crashes if the user configures out the AudioTeeObject.
  • Added an option so that the AudioTeeObject is by default not used, and can be used by setting an option.
  • Added Allow users to set/unset the AudioTeeObject thresholds both from SpeechRunner and from their code.
  • Allow the voice level threshold and file in/out formats (for the phone) to be controlled via options and programmatically. This change specifically supports older Dialogic boards and some modems, where the "supported formats" query returns incorrect results so the format must be forced to a specific value, at least for now.
  • Improved reporting to help user understand what to do if their telephony device setup fails.
  • Tell user why format/stream creation failed.
  • Provide DoNotAnswer property to allow users to control whether SpeechPlayer will answer the phone at all. SpeechPlayer can be set to not answer the phone anymore via a program, so that a program can honor a request, say over the phone, to stop answering the phone. That way if SpeechPlayer programs are causing a problem, there is a remote way to disengage them.  DoNotAnswer applies not just to AutoAnswer, but to all calls.
  • Fixed PRN 287:  Before changing the telephony device, make sure telephony will be initialized. The first initialization (on the IBM laptop) failed preventing any further change to telephony.
  • Provide a property to see who's got control of the phone line.
  • Only answer the phone if an appropriate client is waiting for a connection.
  • Don't answer the phone if we are in the middle of testing.
  • Hang up the phone when we're done. In AutoAnswer we were leaving the connection going, causing the line to be locked until SpeechPlayer is killed. Naturally this annoyed anyone who want to call or get a call.
  • Unify the call id number for simphone and tapi. It turns out that TAPI can get complicated about how it sets up telephone input, so let the call id number reflect calls to Telephone_Connect, and return 0 if the most recent event is a telephone_disconnect.
  • Added "View Log" to the SpeechPlayer menu available from the system tray.
     

SpeechRunner

  • Add support for ignoring selected lines that would otherwise be matched as highlights. This is required because some lines are produced asynchronously and therefore are susceptible to race conditions.
  • Add directive support for control designation, mirroring the directives for LocateControl:  [DownFrom], [UpFrom], [RightFrom], [LeftFrom], and [On].
  • Add the [@x,y] notation for control directives.
  • Adapted the command templates for C#.
  • Allow arbitrary blanks between the SPEAK> and the phrase in AwaitSpeak.
  • Clean up after a simulation from an SQR. We especially must honor the ForceTermination.
  • Allow users to use relative path names for sqr args. The file names will default to the same dir as the sqr.
  • A command line of the form "cmd: //comment" caused a complaint that there was a phantom arg. This fix eliminates an empty arg in this case.
     

Samples

  • Added complete set of samples for C#.
  • Made at least one sqt test available in each VUI sample program, to serve as a use case as well as a test.
  • Supplied Speech Utilities code, so users don't have to write their own common glue code.
  • The samples have been upgraded and further integrated into tutorials.
  • Added samples to illustrate Foreign Window use in each language.
  • Added samples to illustrate control of What Can You Say use in each language.
  • Added samples to illustrate Speak prefix for SPEAKFLAGS use in each language.

Release Notes for Version 3.0.2 (October 28, 2002)

· Fixed PRN 286: Speech Collector: "Record" doesn't start recording

· Fixed PRN 242: Menu text is missing on Windows NT.

Release Notes for Version 3.0.1 (October 1, 2002)

SpeechStudio Grammar Editor

  • The grammar compiler used to be included with the grammar editor. Now the grammar compiler exists as a standalone program, compile.exe. Visual Basic projects are unaffected; however, this change affects how grammar files are compiled in Visual C++ projects. The changes to Visual C++ projects are described in the Quick Start section of the SpeechStudio Help on-line manual.
  • When grammars are compiled as part of building a Visual Basic project, a new set of dialogs help ease the process of fixing and recompiling the grammar files.
  • When grammars are compiled with custom build rules in Visual C++ projects, any errors or  messages from the grammar compiler are directed to the Visual Studio log window.
  • The "Find in File" command is only implemented to search grammars in a project with case sensitive or insensitive searches. In particular, regular expressions are not yet implemented.
  • SpeechStudio has a /display=file option to initialize the contents of the log window, for example, with error messages from a compilation.
  • For Visual Basic projects, SpeechStudio grammars are no longer stored in resource DLLs. The grammars are now compiled into a new file, SpeechStudio.dat, which is loaded at run-time. Visual Basic users no longer have to load the resource compiler, nor are there problems caused by read-only resource files.
  • For Visual C++ projects, SpeechStudio grammars are no longer included in the resource script (.rc file). The grammars are now compiled into a new file, SpeechStudio.dat, which is loaded at run-time. This change simplifies project setup and fixes a problem where Visual Studio would not recompile grammar files.
  • New built-in methods include LogText and StopSpeaking.

SpeechStudio Voice Control

  • The SpeechStudio Voice Control API has been upgraded in many ways. The control is documented in the SpeechStudio on-line help file in the section Programming Reference under SpeechStudio Control API Reference. The voice control encapsulates most of the functionality available for an application program.
  • A new call, Init(), is used to associate a controlling window with each instance of the voice control as well as to provide a search path for the SpeechStudio.dat file. Init() must be called before any other API commands are called.
  • Start() now accepts just the grammar name. It no longer includes a reference to an hWnd, which is now set by the Init() call.
  • A new call, StartEX(), is used to override the search path set by Init().
  • VBStart() has been replaced by Start(). The dll parameter to VBStart() has been moved to Init() and StartEx().
  • The name of the event associated with audio and text-to-speech is now called Speaking.
  • The event (and callback) GetPhraseList has been replaced by the combination of the new Refresh event and phrase maps. This new event provides better control over dynamic grammar generation, in particular it supports initialization, caching, and optimization.
  • Literal strings with colons (":") passed to action routines used to cause incorrect arguments to be passed.
  • The control is now maintaining grammars on a one-to-one basis with the application's grammars. In the past, the control built a single big grammar from all of the app's grammars. Now, each app grammar is loaded, activated, deactivated, and unload separately.
  • The internal command Deactivate() has been replaced by Deactivate(grammarName). This change improves the performance of SpeechPlayer because grammars are now smaller and dynamic grammars are evaluated independently of other grammars in the system.
  • SpeechStudio.dat files have a redistribution key to prevent redistribution of programs created with evaluation copies of SpeechStudio Suite.
  • SetProperty can be used to change the current speaker profile.
  • The Speaking type was updated for C++ programs.
  • The control creates an intermediate window to handle COM events. This window's Class was not being correctly
    removed. This was causing VB programs to get an exception the 2nd time they ran under the VB IDE (run it once and kill the program (click on the x in it's upper right corner) then run it again).
  • The SpeakWav parameter StreamID has been renamed Cookie to reflect that its really not connected to the stream but a chunk of data for the user.
  • SpeakWav now correctly maintains the StreamId parameter.
  • The Speaking event now returns the name of the wav file in question. The old version returned an integer that needed to be converted to a file name.
  • The Dictation API can be connected to an edit control or text box. When an edit control or text box gets control, the Dictation API pops up a small text box to track dictation. When a phrase is recognized, the Dictating event is fired and the application must deal with the text.
  • The Dictation event now returns the hWnd of its associated edit control or text box.
  • The dictation event now returns a list of dictation alternatives.
  • A new method, UseAlternate(), is used to inform the speech engine which dictation alternative to accept as correct (for training).
  • A set of methods and events for Telephony has been added to the control.
  • A crash on exit has been fixed.
  • Several unused properties have been removed.
  • Most parameter names have been changed to be more like Visual Basic names.
  • The C++ event mechanism, sending Windows messages, has been simplified. The number of messages has been reduced from four to one, and the message parameters have been changed. The wParam is the event type. The lParam is a pointer to structure containing information about the event.
  • For C++ programs, a new include file, SpeechStudio3.h, is available. It is very similar to the definitions that you would see with #import, except the types are more friendly and it does not throw exceptions.
  • For phrase maps, the parameters and documentation now use Spoken and Written instead of Key and Value.
  • The audio recording interface now uses the RecordingMode property instead of start and stop recording.
  • Removed SQZGrammar.h which was used to compile C++ rc2 files. It is no longer needed.
  • Changed the name of SpeechStudioControl.dll to SpeechStudio3.dll.
  • Characters that do not have virtual key codes, like '>' and '<', were not handled correctly (because they are shifted versions of other virtual keys). This new code handles these cases using the built-in Windows functions.
  • Implement MapClear(), so users can rebuild maps without having to remember what used to be in them.

SpeechPlayer

  • Add telephone "auto answer" capability to SpeechPlayer.
  • Improve behavior in cases of: no SAPI, no SR engines, no TTS engines.
  • Improve options dialog and logging for the case where there are no supported TAPI devices.
  • Improve telephony logging and error reporting.
  • Move "View Log" and "View Summary" menu items from File to Tools.
  • Record audio for dictation results.
  • Reduce unnecessary redrawing.
  • Grammars are created in the deactive state, and then enabled or disabled according to the muting/input state.
  • Changes to allow SpeechPlayer to work with a half-duplex modem.
  • Change "hot" image for the microphone-mute toolbar button.
  • Double-clicking on the icon in the system tray no longer leaves a bunch of junk in the system tray.
  • The prompts window now displays its horizontal scroll bar only when needed.
  • The buttons that control the scrolling of the prompts window now react much faster. The auto-scroll is faster and smoother as well.
  • Fix two confirmation bugs: confirmation button panel was not shown immediately when going into confirmation mode, and a freed confirmation grammar could be referenced if the client was destroyed before completing confirmation.
  • Default to using the inproc recognizer for audio stream recording.
  • Disable mute controls while running a test.
  • Work around SAPI compilation (gc.exe) bug passing properties.
  • Empty rules cause the SAPI XML grammar compiler to fail with a non-specific error so now the XML grammar is created empty.
  • Draw volume bars with a dimmer color when not listening.
  • Do not shorten phrase lists that are directly reached from <start> as stand-alone lists.
  • If the user clicks the microphone button, the input device is reset to the microphone.
  • Add an internal routine Canonical() to clean up recognized text, for example removing doubled spaces.
  • SAPI, by default, loads grammars as active. SpeechPlayer wants grammars to start inactive. SpeechPlayer uses grammar state to do muting and top-level rule state to show active & inactive grammars. In order to create the grammar without a window of activity, and in keeping with SpeechPlayer's usage model, we create a grammar like this:
    1) Create grammar.
    2) Set grammar to inactive.
    3) Load grammar (rules)
    4) Set rules to inactive
    5) Set grammar to active
  • Add "StartInTray" option for keeping SpeechPlayer hidden in the system tray.
  • Implement pop-to-top, pop-on-confirm display modes.
  • Consolidate Autosize and Show/Hide Prompt options into one "prompt mode" option: show, hide, or auto-show.
  • Add an easy way to restore the default logging levels: the "Default Levels" button on the Logging tab of the Options dialog.
  • Add telephony BlindTransfer method.
  • Add DTMF detection support.
  • Add context help strings for Options dialog controls.
  • Add logging to track how many grammars are loaded into the engine.
  • Reflect changes in the TTS options immediately.
  • Add a simulated telephone interface, SimPhone. It acts like a telephony device, sending the same events but using the microphone.
  • Implement full-time audio stream recording.
  • Better updating of GUI state.

SpeechRunner

  • Add MultiprocessTest command to allow tests to control multiple processes instead of just the process under test.
  • Diff patterns now keep "ReportTestError" messages in the filtered log file by default.
  • If the right panel of a diff window is empty, expanding and hiding differences caused a crash.
  • Fix bug: the SpeechRunner tools weren't being hidden correctly in batch mode.
  • Update the MRU list for Phrase Libraries to include those libraries encountered in OpenPhraseLib commands.
  • The regular expression compiler now reports bad patterns.
  • Check that a regexp pattern is well-formed, and complain if it isn't.
  • If the user is informed of a simulation error, ask if s/he wants to proceed If the user quits a test, the problem (e.g. await window) may conclude after the tester has disconnected. No longer asserts.
  • Implement basic telephone simulation support for SpeechRunner: ring, caller id, connect, dtmf and disconnect commands.
  • Let the user know that the simulation is completed.
  • Lock in the order of speak events for golden logs.
  • Add logger message for test start up.
  • Rename the Insert Say/Paraphrase menu items to remind the user that they use side-by-side simulation.
  • Add new insert commands for Say/Paraphrase so that side-by-side simulation is not required.
  • Add the template support for the Speak and the DTMF commands.
  • Show marker-type errors in the log window.
  • Speak() and SpeakWav() now have distinctive log messages, which are considered highlights.
  • Standardize error messages so they can be displayed nicely by SpeechRunner.
  • More errors are reported through SpeechRunner instead of stopping SpeechPlayer with message boxes.
  • SpeechRunner now waits for a short duration for a new grammar to become active, reducing the need for too many calls to AwailtWindowListening.
  • Add signal value for "too busy to simulate now".
  • SpeechRunner diff: Add support for coloring the log bar lines.
  • Add the PhraseLibSpeaker command so that SpeechRunner can use phrase lib files recorded under a name that is different from the current engine name.
  • Allow apostrophe's in phrases.
  • Allow more time if we are waiting for a timeout operation.
  • Allow the user to provide defaults for choosing the exe name or the window name.
  • Change Await* logging so that the command is shown as PENDING> until it either succeeds or fails. That way it will be synchronized with the APP messages.
  • Close the log file after a simulation, so the user can use the diff operation on it.

Grammars

  • Add checks for floating point as grouped numbers.
  • Add a time and duration grammar.
  • Add ordinals.
  • Add small_float and tiny_float.
  • Allow grouped military letters.
  • Allow negative exponents.
  • Disallow the word "decimal" in front of "point".
  • Allow a simple number without a decimal point.
  • Allow input of numbers in dollar or decimal format.

Samples

  • Add recorder samples.
  • Add telephone samples.
  • Add dictation samples.
  • Add the Problem Call application.
  • Fix mouse up event handler in Scribble.
  • Change the name of Scribble's "Pen Widths.grm" to "PenWidths.grm".

Other changes

  • Add licensing to Phrase Explorer.
  • Add licensing to SpeechCollector.
  • Change the GUIDs and version numbers to reflect version 3.
  • Change the name of the eula for Suite to Suite. Move the release notes to the top-level directory.
  • The release notes & installation instructions are combined.
  • Converted the logging directory to .../SpeechStudio from .../SQZ.
  • Break out the tutorials and collect them in a tutorial directory.
  • No longer distributing ComVector.h with our product.

Release Notes for Version 3.0.0 (September 9, 2002)

  • Beta release

Release Notes for Version 2.0.5 (March 19, 2002)

· New Feature: Added a SpeechPlayer registry entry, StartInTray, that tells SpeechPlayer not to bring its window up.

· Bug fix: SpeechPlayer changed the message numbers of all messages for which it listens into a safer range. Other (non-malicious) programs should not be able to interfere with SpeechPlayer now.

· Bug fix: An intermediate window class was not being correctly removed, causing an error in a rare case.

· New Feature: SpeechPlayer redistributable no longer requires speech recognition components if the application is only doing TTS.

· New Feature: SpeechPlayer no longer complains if SpeechRecognition is not installed in the case where the application is only doing TTS.

Release Notes for Version 2.0.1 (November 21, 2001)

Version 2.0.1 brings the Visual C++ and Visual Basic support in sync and creates an Enterprise product containing support for both.

· Changed the C++ Sample name "SQZedCalc" to "SQZCalc with VUI"

· Bug fix: recording audio and confirmation didn't mix, because the audio for the recognition-to-be-confirmed was not being kept around.

· The help for SpeechStudio has been combined into one manual with both VB and VC++ information. The same is true for SpeechRunner.

· SpeechStudio is now licensed by language (VB or VC++ or both).

· Color scribble example for VC++ can now tune colors on the fly.

· Bug Fix: Dictation could only be started by a recognition. That is, if you pressed a button to do dictation, it didn't work.

· The company name has changed from "SQZ Inc." to "SpeechStudio Inc." To reflect this change we have changed the titlebars, icons, program names, and most other user-visible locations where SQZ used to appear.

· The SpeechStudio product is now installed in C:\Program Files\SpeechStudio instead of C:\Program Files\SQZ

· In SpeechRunner if the path to the .exe file is "./", then "." is defined to be the location of the current .sqt file.

· A better Voice user Interface has been added to the VB calculator example to create Samples/VB/vbCalc with better VUI. In particular the spelling, define, options, and spellcheck dialogs have been unified.

· In SpeechRunner, allow the use of the Static control "neighbor" to identify an Edit box for the DoSetText and DoEditBoxCheck commands.

· Include the SpeechPlayer redistributable as part of the product. See the Redist folder in the product (there is a readme file there as well).

· Changed Calculator so that variables and functions are now distinct.

· The VB add-in now watches for events that require the SpeechStudio.dll to be up-to-date, events such as execute, file | make, etc. If the DLL is not up-to-date, then it invokes SpeechStudio to bring it up to date.

· SpeechPlayer will now always come to the top when confirming, even if it is iconified.

· Fixed some SpeechPlayer redrawing bugs when going from minimized to normal size.

Release Notes for Version 2.0.0 (October 21, 2001)

Version 2.0.0 is a Visual Basic only release.

SpeechStudio Suite for Visual Basic

Improvements to the SpeechStudio Resource Manager Add-In for Visual Basic include:

· The maintenance dialog, brought up by pressing the eyeglasses button, allows for the selection of the SQZ project file and, if the VB project file is not read-only, this selection will be recorded so the user won't be prompted in the future.

· In a VB project that contains SpeechStudio voice recognition components, prior to executing a VB program from the IDE or building a .exe (i.e., file | make...), the add-In will check that the SpeechStudio.dll is up-to-date and, if not, will bring it up to date by invoking SpeechStudio to compile the .sqz file.

· Better error reporting

The dictation component of SpeechRunner has been improved to put blanks between the groups of words as they are recognized.

There is a new StopSpeaking() method in the SpeechStudio control which terminates audio output immediately (TTS or .wav output).

Release Notes for Version 1.7.8 (September 5, 2001)

Version 1.7.8 is a minor upgrade to 1.7.7 to add example SpeechStudio.dll files into two sample directories in Samples\VB\: Events and vbCalc with VUI.

The SpeechStudio help document has also been upgraded.

Release Notes for Version 1.7.7 (July 19, 2001)

SpeechStudio Suite for Visual Basic

Version 1.7.7 is the beta release of SpeechStudio Suite for Visual Basic.
The non-VB-specific bug fixes and feature changes for the product components will be documented in the full product release.

· A new component called The VB Add-In, is part of the VB product. Installation of the product will automatically enable the Add-In's toolbar, called the SpeechStudio toolbar, in Visual Basic. Use of the buttons on the toolbar is described in the SpeechStudio Help.

· SpeechStudio can now get it's resource descriptions from an .xml file, which is what the VB Add-In creates.

Known problems in the VB Beta

· None of the Updown control actions in a grammar will work.

· The actions of the slider control (aka msctls_trackbar) work, however they do not cause the event procedures in a VB program to be triggered.

· None of the scrolling actions for list and combo boxes are working.

You can work around most problems such as these by calling the FireRecognized action in the grammar instead of the control-specific action. Then just perform the desired action in the Recognized event procedure.

· There is no way in the beta to voice activate the three system menu buttons usually appearing in the upper right hand corner of the window (minimize, maximize, and quit). As described above, you can use FireRecognized to work around this problem if desirable.

· SpeechStudio Inc. just recently changed its name from SQZ Inc. There are still some vestige references to SQZ in our help documents, error messages, etc. For those new to SpeechStudio Inc., this message explains where that "SQZ" you see comes from.

Release Notes for Version 1.7.1 (April 11, 2001)

SpeechStudio Suite C++

· Saving a read-only file sometimes resulted in a error message that displayed the "Help About" message.

· Log files for all SpeechStudio programs are now written to the "SQZ" folder in the system's Application Data folder, instead of the root directory.

· Sped up string handling for XML parsing.

· Support hyphenated words like forty-one. Before the hyphenation was treated as a confidence operator; now it is a hyphen if an ID character immediately follows it. Numbers must now be given in hyphenated form.

· Improved SpeechRunner and SpeechPlayer cooperation in cleaning up killed processes.

· Improved handling of audio recording. Audio recording now works correctly when multiple clients are recording simultaneously, in the same or different directories.

· Copyright messages have been updated to include the year 2001.

· Several additional predefined grammars are now available in addition to <integer/>. They are natural, digit, digits2, and digits3.

SpeechStudio C++

· In batch mode, SpeechStudio did not write error message or return a bad error status. Now in batch mode, errors are written to standard output and the number of errors is returned in the exit status.

· Implemented Save As... for grammar files.

· SpeechStudio would, in rare cases, crash immediately after briefly flashing its splashscreen.

· Changed the SpeechStudio grammar-attribute "disable" to be "disabled".

· In SpeechStudio grammars, add "true" and "false" as Boolean values, joining "yes", "no", "1", and "0".

· Added a new tutorial program, "Color Scribble with Playback".

· File | Refresh now marks the project file as written when resources attached to the project are changed.

· File formats are automatically upgraded from version 1.5 to version 1.7.

· SpeechStudio now supports grammars that are not associated with resources. The grammars, which may be started and stopped from anywhere in an application, can communicate with the application using GetPhraseList and FireRecognized.

· SpeechStudio now associates grammar names, not grammar files, with resources. Grammar files can be shared between grammar names. Multiple grammar names can be associated with a single resource. Taken together, these new features combine to allow reuse of primitive functions, such as common control dialogs.

· Grammars can now access the value dictated to the wildcard patterns "*" (dictation) and "..." (junk).

· Reconciliation has been simplified so that only grammar files must be reconciled, and then only when the resources associated with the grammar file undergo a significant change.

SpeechPlayer

· Internal debugging routines no longer fail with an assertion error if the output message is too long.

· SpeechPlayer now uses the shared recognizer when possible, so that SpeechStudio-enabled programs can run alongside Office XP and other programs that use SAPI 5 directly.

· When the current SpeechPlayer client is killed abnormally (for example, by the Win32 API function TerminateProcess), stop listening immediately.

· SpeechPlayer has more efficient grammar loading. Grammars are only written to a temporary file if an error occurs, or if the user has requested debugging information by setting a high logging level.

· SpeechPlayer has improved logging for speech recognition engine events and grammar state changes.

· SpeechPlayer sometimes terminated a test immediately upon starting it. This rare event occurred when a test was just terminated unnaturally and left timer routines running, so that these stale timers triggered during the new test.

· Put in score tuning for the Microsoft SAPI 5 engine. Numeric scores between -100 and 102 can now be used for judging confidence for the SAPI 5 engines. Values less than zero are false recognitions. Values above 100 indicate that the result was simulated or forced.

· The Properties | Set Logging Levels dialog now supports a button to reset to defaults, resetting to the original logging levels.

· Added menu item File | View Log to quickly access the most recent part of the SpeechPlayer log.

· The Microsoft SAPI 5 (or engine) has a bug returning properties for grammars that use multiword string literals as default VALSTRs. Occasionally only the last literal word is returned as the VALSTR property. Work around this problem by simulating the correct result construction.

· If a SpeechPlayer client starts without registering an hWnd, an addition message may be displayed later when the client registers its hWnd.

SpeechStudio Control

· The control is incompatible with version 1.5.

· Fixed built-in actions affecting Listbox selection: for a listbox with multiple selection style (LBS_MULTIPLESEL), speaking the name of a list item that is selected will unselect it.

· Built-in actions affecting Listbox selection always send change notifications, even if the actual selection does not change.

· The API Speak method was limited to a short, fixed, number of characters to speak. Now this number is unlimited; however, speaking cannot be interrupted so speaking in short phrases is preferred.

· Built-in Actions now have limited support for the Windows common control ListView.

· SpeechStudio Control maps, which are used to handle abbreviations or dynamic wordlists, are no longer associated with controls. Instead, maps have unique identifiers. These identifiers are used throughout map-related methods (such as GetMappedWordlist and MapInsert).

· A new built-in action, GetPhraseList, is used to dynamically create grammar elements. When the grammar is evaluated, GetPhraseList caused the control to send the GetPhraseList event (or Windows message) to the application.

· SpeechStudio Control signatures are no longer used. The Signature property has been removed.

· The built-in action FireRecognized now takes a variable number of arguments.

· The Recognized event (or Windows message) has a new parameter profile. In particular, the event now has an associated id and score. If the phrase is being recorded, the id number of the recording is set. The parameters to FireRecognized are passed in a single SafeArray.

· Built-in actions which appear in quoted strings (for example, GetPhraseList), will implicitly and automatically have their first argument quoted to avoid the XML quoting syntax, e.g. &quot;id&quot;.

· The control no longer tries to get dictation text from a null hWnd.

· The Recorded event (or Windows message) has a new parameters profile.

· The Recorded event (or Windows message) for a false recognition is sent to all clients that have registered to receive false recognitions.

· The built-in action Dictate that was associated with edit (and richedit) windows has been replaced by StartDictation.

· The API implements a new interface for examining and changing SAPI engine properties, SetProperty and GetProperty.

· The API methods Speak and SpeakWav now take an addition parameter, a stream id. The stream id is communicated back to the application with the Speaking event (or Windows message).

· The API methods and events are now suitable for automation (and compatible with Visual Basic).

· The control always sends the Dictated event (or Windows message) after dictation, even if the id is omitted.

SpeechRunner C++

· SpeechRunner now waits until the test program's log file has been closed before checking the test results. If an error occurs during the check, it now reports more useful information.

· SpeechRunner could sometimes not start a test without waiting for a client: SpeechPlayer complained that the engine had no grammar.

· Introduced the SetParameter: StartWithoutClient switch to tell SpeechPlayer that we're starting the test even though the engine or process may not be ready, so it doesn't complain. This parameter is set automatically by SpeechRunner, under normal circumstances, when "Start Without Client" is selected in the Run dialog.

· In some cases quotes from incoming string arguments were not removed.

· Introduced the AwaitLogText command for SQT files. This command allows SpeechPlayer to monitor log messages as they arrive from the app, and delay the next test action until a log message matches a given pattern.

· Show Differences did not work correctly if the output line had a "|" character in it.

· Standardized the "Test File Completed" message so that it comes out whenever the test is stopped, as long as the SQT file has been completely processed.

· SpeechRunner no longer will complain about missing phrase libraries when simulating.

· In rare cases, an infinite loop of repeated errors could occur if an SQT test was edited to name bad phrase libraries.

· Implemented SpeechRunner DoSetText command, so that a user can push specific text into a control from an SQT test.

Release Notes for Version 1.5.1 (December 14, 2000)

· Replaced the "spreadsheet" grammar entry display with the XML smart editor.

· Minor changes and bug-fixes in SpeechStudio.

Release Notes for Version 1.0 (November 7, 2000)

· Major changes switching from SAPI 4.0a to SAPI 5.

· The documentation was improved. Tutorials were enhanced for SpeechStudio and SpeechRunner.

· Introduced SpeechCollector and Phrase Explorer for test suite management

Release Notes for Version 0.9 Beta (May 5, 2000)

· Introduced SpeechStudio Suite for SAPI 4.0a
© 2001-2005 SpeechStudio Inc.  All Rights Reserved.

Privacy Policy