This document describes system requirements, installation procedures, and last-minute known issues for Microsoft® Office Communications Server 2007 Speech Server.
Updated release notes can be found at http://go.microsoft.com/fwlink/?LinkId=94652.
This Quick Install Guide for Developers is intended for voice response application developers who want to get Speech Server Development Tools and Server Components up and running as quickly as possible. This guide assumes a single-server topology and provides no scalability instructions.
The instructions that follow are not intended to guide installation for enterprise deployment or voice response application deployment. For enterprise installation instructions, install the Documentation component first and then see the Help topic "Installing Speech Server". For information on voice response application deployment, see the Help topic "Speech Application Deployment".
The Speech Server installer enables you to install the following Speech Server components:
Note Speech Server Analytics and Tuning Studio (included in the Development Tools component) supports application tuning and analysis with English (United States) operating systems using English (United States) locales only. However, within that environment you can analyze data collected from voice response applications running in any language that Speech Server supports.
Important Planned improvements in Speech Server tuning tools features and functionality have resulted in changes to the tuning database schema in each successive release of Speech Server. As a result, the tuning tools for each successive release of Speech Server are incompatible with the databases that were created using the tuning tools in previous releases of Speech Server. For example, you will not be able to use Speech Server tuning tools with databases created using the Speech Server Beta 1 or Speech Server Beta Refresh tools. For all databases created before this release, maintain a copy of the tools with which you created the tuning databases if you need to retain the databases for business purposes.
If you do not want to install all Speech Server components, you can install only the Documentation and Administrative Tools by selecting these components and clearing all other component check boxes on the Component Installation page of the Installation Wizard. You can use the Administrative Tools to administer Speech Server deployments on local computers and remote computers.
The computer on which you install Speech Server must meet the following hardware requirements.
| Feature | Requirement for Speech Server Installation |
|---|---|
| Processor | Pentium 4 class running at 2.5 gigahertz (GHz) or faster |
| Number of Processors | 1 (2 recommended) |
| RAM | 1 GB (4 GB recommended) |
| DVD-ROM Drive | Required only if installing from DVD rather than from a network |
| Network Interface Card (NIC) | Required |
| Free hard disk space (for installation) | NTFS-formatted local partition with 3.5 GB of available hard disk space |
| Free hard disk space (for logging) | 5 GB minimum for Speech Server with Development Tools; 20 GB minimum for Speech Server with other components. Requirement depends on the amount of log data you want to store on the computer. |
The following are additional requirements and recommendations for computers that will be used to develop voice response applications.
| Feature | Requirement or Recommendation for Speech Application Development |
|---|---|
| Video | Super VGA (800 x 600) or higher-resolution adapter and monitor |
| Keyboard and Mouse | Required |
| Microphone | Optimal performance is achieved with a high quality close-talk (headset) microphone. A universal serial bus (USB) microphone with gain adjustment support is recommended |
| Audio Output Device | Recommended for application development |
You must be a member of the Administrators group on the target computer to install Speech Server and all prerequisite software. To verify that you are an Administrator, see the User Accounts Control Panel on your computer. If you are not an Administrator, see your network administrator.
You must be a member of the Administrators group to open the sample and reference applications, or solution files.
Administrative permissions are required to develop voice response applications and to use the Re-recognizer tuning tool. However, administrative permissions are not required to use the other Analysis and Tuning tools or to view tuning reports.
Speech Server sets permission levels on the computers where it is installed. When Speech Server is installed on a computer running Windows Server 2003, only administrators and power users (not users or guests) are able to log on interactively. However, when Speech Server is installed on a computer running Windows XP or Windows Vista, users in both the Admin group and the Interactive Users group are allowed to log on interactively. For more information, see the Help topic "Logon Permissions for the Computer Running Speech Server".
All components listed in Speech Server Components are supported on one of the following operating systems:
| Server Operating Systems | Client Operating Systems |
|---|---|
|
|
The prerequisites required for installing Speech Server depend on the Speech Server components that you choose to install. The following table illustrates the prerequisites for these components and the recommended order in which the prerequisites should be installed. Click the name of the prerequisite to view installation procedures and notes.
After installing prerequisite software, install the latest service packs and all critical updates for your operating system from Windows Update.
| Prerequisites | Server Components | Development Tools | Documentation | Administrative Tools | Data Processing Tools |
|---|---|---|---|---|---|
| IIS and MSMQ | Required | Required | − | − | − |
| Microsoft .NET Framework 3.0 Redistributable Package | Required | Required | Required | Required | Required |
| MSXML 6.0.3883 | Required | Required | − | − | − |
| Visual Studio 2005 SP1 (and for Windows Vista only, Visual Studio 2005 SP1 Update for Windows Vista) | − | One of the following Editions is Required:
| − | − | − |
| Microsoft Visual Studio 2005 Extensions for Windows Workflow Foundation | − | Required | − | − | − |
| Hotfix for Microsoft .NET Framework 2.0 (KB917833) (Windows Server 2003 and Windows XP only) | Required | Required | − | − | − |
| Integrated Windows Authentication enabled | Required | Required | − | − | − |
| Microsoft Document Explorer (see Additional Prerequisites) | − | − | Required | − | − |
| Microsoft Visual C++ 2005 SP1 Redistributable (see Additional Prerequisites) | Required | Required | − | Required | Required |
| Microsoft Office Communications Server 2007, Unified Communications Managed API (UCMA) v1.0 (see Additional Prerequisites) | Required | Required | − | − | Required |
Following are procedures for installing Speech Server prerequisites and additional notes about the prerequisites.
On Windows Server 2003
Note Selecting ASP.NET on 32-Bit systems or Internet Information Services (IIS) on 64-Bit systems causes Windows to automatically select all of the components of IIS that are required for Speech Server.
On Windows XP Professional
On Windows Vista
Note When you enable the features indicated in the table by boldfaced, blue font, Vista will automatically enable additional, dependent features. These additional features are indicated in the table by solid, black discs.
|
Note The features that this procedure directs you to enable together with the dependent features that Vista automatically enables are the minimum set of features that are required to install Speech Server on computers running Vista.
To install Microsoft .NET Framework 3.0 Redistributable Package
On Windows Server 2003 and Windows XP
Important Speech Server does not support pre-release versions of .NET Framework 3.0. If you have a pre-release version installed, you must first uninstall the pre-release version and then install the released version.
Note The final dialog screen of the .NET Framework 3.0 Installation Wizard directs you to install Update for Windows Communication Foundation (KB912817). If you accidentally close the .NET Framework 3.0 Installation Wizard before installing this update, click the link provided in this note, and then download and install the version of this update that is appropriate for your computer's architecture.
On Windows Vista
To install Microsoft XML 6.0.3883
If you do not have a copy of Visual Studio 2005 SP1, you can download or order an evaluation copy of Professional Edition or Team Suite Edition from the Microsoft Visual Studio MSDN Evaluation Center.
Note Speech Server tuning tools require Microsoft SQL Server™. If you plan to use Speech Server tuning tools and you do not already have SQL Server 2000 or SQL Server 2005 installed on your computer, select the option to install SQL Express during the Visual Studio 2005 SP1 installation process. Speech Server Reporting Services (RDL) reports require SQL Server 2005 (Standard Edition or later).
If you already have Visual Studio 2005, you can download Visual Studio 2005 Service Pack 1 for free from the Microsoft Developer Center.
Important If you are installing Speech Server on a computer running Vista, in addition to installing Visual Studio 2005 SP1, you must also install Visual Studio 2005 Service Pack 1 Update for Windows Vista. The Visual Studio 2005 Service Pack 1 Update for Windows Vista addresses areas of Visual Studio impacted by Vista enhancements.
To download Visual Studio 2005 Service Pack 1
Note Visual Studio 2005 Service Pack 1 is a large program that can take a long time to install. As a rule, allow as much time to install Service Pack 1 as you did to install Visual Studio 2005.
To download Visual Studio 2005 Service Pack 1 Update for Windows Vista
To install Visual Studio 2005 extensions for .NET Framework 3.0 (Windows Workflow Foundation)
Speech Server does not support pre-release versions of Visual Studio 2005 Extensions. If you have a pre-release version installed, you must first uninstall the pre-release version and then install the released version. To determine what version of Visual Studio 2005 Extensions is installed on your computer, do the following:
To install the Hotfix for Microsoft .NET Framework 2.0 (KB917833)
Note This hotfix is not required on Windows Vista systems.
To enable Integrated Windows authentication
On Windows Server 2003 and Windows XP
On Windows Vista
Additional Prerequisites
The Speech Server installer automatically installs the following three additional prerequisites on your computer as necessary when you run the Speech Server Installation Wizard, or when you perform an unattended (silent) installation using the /ADDLOCAL option to specify any of the components that require them:
If these additional prerequisites are not installed on your computer, performing a silent installation without using the /ADDLOCAL option can result in failure or an unsupported configuration.
If you need to perform a silent installation without using the /ADDLOCAL option, then you must first install these additional prerequisites. Use the installers provided in the Prerequisites directory at the root of the Speech Server installation disc to install the prerequisites in the order in which they appear in the previous list.
Note On 64-Bit operating systems, both the 32-Bit and 64-Bit versions of Microsoft Visual C++ 2005 SP1 Redistributable and UCMA v1.0 are required.
For information on how to perform a silent installation using the /ADDLOCAL option, run the Speech Server Installation Wizard to install only the documentation, and then see the topic "How to: Run a Silent Installation".
Before installing Speech Server, deactivate any real-time virus monitoring software that is running on the target computer. Once installation is complete, reactivate real-time virus monitoring. In addition, close any running programs before installing or uninstalling Speech Server.
To install Speech Server
Important The final screen of the Speech Server Installation Wizard prompts you to install a Speech Server Language Pack from either the installation DVD or from the Web. Speech Server does not support user interaction in any specific language by default. You must install at least one language pack in order to obtain support for user interaction.
Uninstalling Speech Server removes all components listed in Speech Server Components.
Important Before uninstalling Speech Server, close any running programs.
To uninstall Speech Server
| Problem: | |
| Speech Server requires the ASP.NET Windows and .NET Extensibility features to be enabled. However, when I upgraded my Windows XP computer to Vista, I was unable to install these features. When I tried to install ASP.NET, I received the following error: | |
| Solution: | |
To resolve this issue, follow these steps:
Note To reinstall the ASP.NET on Windows Vista, open the Windows Features window by following the first three steps of the previously listed procedure describing how to install IIS and MSMQ on Windows Vista, and then enable the ASP.NET feature. For additional information, see the following KB article: http://support.microsoft.com/kb/929772/en-us. |
|
|
|
|
| Problem: | |
| When the Speech Server installer is run on computers running Windows Vista, it will allow you to install Speech Server even if you have not enabled all of the IIS features listed in these Release Notes as prerequisites. As a result, Speech Server might not function as expected. | |
| Solution: | |
| Resolve this problem by ensuring that you enable all required IIS features on Windows Vista listed in the section "To install IIS and MSMQ" of this document before you run the Speech Server installer. | |
|
|
|
| Problem: | |
| When the Speech Server installer is run on computers running Windows Vista, it will allow you to install Speech Server even if you have not enabled all of the MSMQ features listed in these Release Notes as prerequisites. As a result, Speech Server might not function as expected. | |
| Solution: | |
| Resolve this problem by ensuring that you enable all required MSMQ features on Windows Vista listed in the section "To install IIS and MSMQ" of this document before you run the Speech Server installer. | |
|
|
|
| Problem: | |
| When Speech Server is running on Vista, disabling Trace Logging fails. An error is logged to the Windows Event Log and Speech Server continues to log to the ETL file. | |
| Solution: | |
After disabling Trace Logging through the MMC, open a command window on the computer running Speech Server and run the following command:
logman -ets stop "MSSTrace" |
|
|
|
|
| Problem: | |
| I cannot start the Speech Server service after upgrading my computer operating system from Windows XP to Windows Vista. | |
| Solution: | |
| For the security reasons, the Windows Vista upgrade process disables the World Wide Web Publishing Service. This stops the Default Web Site in the Internet Information Services (IIS) Manager. In order to start Speech Server service normally, you must open the Services window of the Microsoft Management Console and enable World Wide Web Publishing Service, and then open Internet Information Services (IIS) Manager and restart the Default Web Site. | |
|
|
|
| Problem: | |
| If the disk drive on which the Speech Server Event Trace Log (ETL) files are being saved becomes full, the logging process stops even though Speech Server can still be actively handling calls. Log data is not recorded until the logging process is manually restarted. | |
| Solution: | |
To resolve this problem, first free up space on the disk drive that is used to save the ETL files and then restart logging.
To free up disk space To prevent this problem from occurring, consider configuring a low disk space alert. For information on how to configure a low disk space alert by using the performance logs and alerts feature in Windows Server 2003, see KB Article 324796. |
|
|
|
|
| Problem: | |
| Installing Microsoft Speech Application SDK (SASDK) 1.1 on top of Speech Server results in a Speech Server failure. This problem is indicated by an error in the event log such as the following: The reason for this is that the SASDK 1.1 TTS engines are incompatible with Office Communications Server 2007. | |
| Solution: | |
| Resolve this problem by uninstalling SASDK 1.1 and then reinstalling the Server Components of Speech Server. | |
|
|
|
| Problem: | |
| I am running Speech Server on Windows Vista. When I try to call my application from an external source, Speech Server does not connect. | |
| Solution: | |
| One possible source of this problem is that the Windows Firewall on Vista may be blocking the connection. To connect, you must create exceptions for Microsoft.SpeechServer.Tap.exe and w3wp.exe. | |
|
|
|
| Problem: | |
| My application successfully accepts calls from a locally hosted VoIP source (softphone, development environment, or TIMC). When I deploy my application to a server other than the VoIP source, it no longer accepts calls. | |
| Solution: | |
One possible source of this problem is that calls are blocked by the IPSec Service running on the server where the application is deployed. In this case, stopping the IPSec Service will resolve the problem. This applies only to topologies in which calls originate from a VoIP server other than the server hosting the application (such as a VoIP gateway or softphone).
Note IPSec normally blocks inbound connections to Speech Server. However, if a Speech Server outbound call was accepted by a gateway, then for several minutes following this, Speech Server will accept calls from that source even if IPSec is on. For deployments hosted on computers running Vista, another possible source of this problem is that the firewall is not configured correctly. On computers running Vista, ensure that you add exceptions to the firewall for the following:
Important Deployments on computers running Vista should be used for testing purposes only because support for Vista is limited to two localhost channels. |
|
|
|
|
| Problem: | |
| If you install Telephony Interface Manager Connector (TIMC) without first installing a Telephony Interface Manager (TIM), Speech Server will not start because the operation of TIMC is dependent on the presence of a TIM. | |
| Solution: | |
| TIMC is intended to be installed with a TIM. If you do not plan to use a TIM, do not install TIMC. | |
|
|
|
| Problem: | |
| When I use MssLogToDatabase on a machine running .NET Framework 2.0, I either receive tabular data stream protocol errors, or the import times out. | |
| Solution: | |
| Resolve this problem by installing the .NET hotfix described at http://support.microsoft.com/kb/913177/en-us on the computer that is running MssLogToDatabase. | |
|
|
|
| Problem: | |
| When I try to play back session audio for which there were no dialog turns, Analytics and Tuning Studio says "The system cannot find the file specified". When I try to export session audio for a session which contains no dialog turns, I am prompted for a folder, but no audio file is saved. | |
| Solution: | |
| Analytics and Tuning Studio is unable to recreate the audio for sessions in which there were no dialog turns. Find sessions which contain dialog turns (for example, for which the turn count value is > 0) and playback or save the audio for those sessions. | |
|
|
|
| Problem: | |
| If I save an Analytics and Tuning Studio query file that specifies a value for the Session Label filter and point it to a different database, the data returned by the query is incorrect. | |
| Solution: | |
| When viewing the latter database, uncheck the Session Label filter, refresh the query, and then re-check the Session Label filters. The correct data should be returned. | |
|
|
|
| Problem: | |
| When I attempt to run a re-recognition cycle with the same name as an existing re-recognition cycle, and select Replace when prompted, I receive the following alert: | |
| Solution: | |
| Give the re-recognition cycle a different name before executing. | |
|
|
|
| Problem: | |
| In Analytics and Tuning Studio, when I try to navigate from a Turn view selection up to a Task or Session view, or from a Task view selection up to a Session view, the navigation does not reflect only my selection, but instead displays all data as selected by the filters. | |
| Solution: | |
| To see containing session or task information for specific turns or tasks, apply an explicit filter to return only the sessions or tasks of interest. | |
|
|
|
| Problem: | |
| I used MssLogToText to convert a log file to text, and now I am examining the log content. However, I do not understand the log schema, and I cannot find documentation about the schema. | |
| Solution: | |
Log schema documentation is provided in a separate HTML document. When Speech Server is installed to the default location, the log schema documentation is located in:
%ProgramFiles%\Microsoft Office Communications Server 2007 Speech Server\doc\EventSchemaForHelpDocs.htm |
|
|
|
|
| Problem: | |
|
Sometimes, after an execution of MssDataPurge, I see rows left behind that should have been deleted and/or I am unable to re-import a log file whose data was previously purged.
This is because in certain cases where trace messages cannot be associated with sessions in the data (for example, because call events are split across import sessions or events are otherwise left "orphaned"), the trace messages cannot be removed by MssDataPurge. |
|
| Solution: | |
| Orphan rows in the Messages table should be manually removed. | |
|
|
|
| Problem: | |
| Using Analytics and Tuning Studio, when I set filters in Turn Details View to view turns that have no TurnName, the size of the TempDB dramatically increases. | |
| Solution: | |
| Analytics and Tuning Studio does not support querying the Turn Details View in an unconstrained fashion. Either filter your query to include turn names or other turn identifiers, or make a selection in the Turn List View and navigate to the Turn Details View with the selection. To recover the disk space consumed by the TempDB , restart SQL Server. | |
|
|
|
| Problem: | |
| When I try to purge recognition cycles from my tuning database using MssDataPurge.exe with the /pre: and /post: arguments, it does not appear to work. | |
| Solution: | |
| As with the other types of data purging, the /pre: and /post: parameters refer to the time of the session, not the time that the re-recognition cycle was conducted. To purge re-recognition cycles based on dates, use /pre: and /post: to specify the time period for the sessions holding the original recognition. | |
|
|
|
| Problem: | |
|
When viewing Speech Server documentation, I encounter broken links in the API topics.
The Speech Server documentation is designed for integration with the MSDN documentation collection. When Speech Server documentation is installed on a computer without MSDN, all links to MSDN topics are non-functional. Most of these links can be fixed by installing the MSDN collection. In addition, the Speech Server product uses components of Windows Workflow Foundation, and consequently, the Speech Server documentation references Workflow documentation. Unfortunately, the Workflow documentation is not yet part of the MSDN collection. After the Workflow Foundation documents have been added to the MSDN collection, it will be possible to fix these links by installing an MSDN collection that contains the Windows Workflow Foundation documentation. |
|
| Solution: | |
| Install the latest version of MSDN to minimize broken links in the Speech Server documentation. | |
|
|
|
| Problem: | |
|
When running the SupervisedTransfer sample application, I get a System.InvalidoperationException when I enter any extra text in addition to a valid name in the askContactName QA, for example, when I enter "John please".
The reason for this is that the askContact QA uses the text of the recognized result directly rather than using semantic values. |
|
| Solution: | |
Resolve this problem by replacing the following block of Incorrect Code with the following Replacement Code:
Incorrect Codeprivate void askContactName_Closed(object sender, EventArgs e) Replacement Codeprivate void askContactName_Closed(object sender, EventArgs e) |
|
|
|
|
| Problem: | |
|
Sometimes while using Analytics and Tuning Studio on a computer running Microsoft Vista, Visual Studio stops responding. This happened once when I was switching from a list view to different view and another time when I tried to play audio.
There is a bug in Visual Studio that causes it to stop responding on computers running Microsoft Vista when the following two conditions occur simultaneously:
|
|
| Solution: | |
| While waiting for a dependent application to start, avoid moving the mouse pointer over an object that has a tooltip until the dependent application finishes starting. Also, before shifting focus between panes or windows, move the mouse pointer away from objects that display a tooltip unless the tooltip is fully rendered. | |
|
|
|
| Problem: | |
|
In Visual Studio while developing workflow or voice application workflow applications, the "Windows Workflow" tab is missing from the toolbox.
This can happen if C++ settings are the applied settings. |
|
| Solution: | |
| Resolve this problem by right-clicking the toolbox and choosing Reset Toolbox from the context menu. | |
|
|
|
| Problem: | |
| Building a Visual Studio solution immediately after having removed all pronunciations from the Custom Application Lexicon (CAL) will cause a build error that prevents the application from running. | |
| Solution: | |
| Resolve this problem by clicking the Build button on the CAL editor user interface before using Visual Studio to build the project or solution. | |
|
|
|
| Problem: | |
| Editing the Custom Application Lexicon (CAL) associated with a grammar in the Conversational Grammar Builder does not cause the dirty flag of the grammar to be set. The grammar needs to be rebuilt to absorb the changes made in the CAL, but the user is not informed of this via the dirty flag. | |
| Solution: | |
| Resolve this problem by rebuilding the Visual Studio Solution or rebuilding the CGB grammar after every CAL editing session. | |
|
|
|
| Problem: | |
|
Using the Custom Application Lexicon Editor to change the pronunciation of a word appears to have no effect on whether or not that word is highlighted as being acoustically similar to other words when I click Check Acoustic Confusability in Simple Grammar Builder.
The reason for this is that during the acoustic confusability test, the Custom Application Lexicon is not used, thus any pronunciation modifications made in that lexicon are not reflected in the test. |
|
| Solution: | |
| There is no workaround for the acoustic confusability test issue. However, when the grammar is used at runtime, the Custom Application Lexicon is in fact employed and pronunciation edits are accurately reflected. | |
|
|
|
| Problem: | |
| I opened a .grxml grammar rule using Grammar Editor. While the rule was open, I changed a property of my desktop display. When I returned to Grammar Editor, the visual layout of the rule was corrupted.
This issue is caused by a bug in a component on which the Grammar Editor depends. Actions that cause a redrawing of the computer monitor's display while a grammar rule is open can trigger this issue, including the following:
The resulting layout corruption can include the following:
|
|
| Solution: | |
| Close and then reopen the rule. | |
|
|
|
| Problem: | |
|
Using Conversational Grammar Builder, when I click Parse to parse my grammar, I get the following error:
My conversational grammar references a keyword (.grxml) grammar.
This error indicates a mismatch between the language of the speech recognition engine that Conversational Grammar Builder is using and the language of a grammar that Conversational Grammar Builder is attempting to load. For example, if the conversational grammar in your English (United States) application references a French (Canada) keyword grammar, parsing the conversational grammars generates this error because Conversational Grammar Builder is attempting to load and parse a French (Canada) keyword grammar using an English (United States) speech recognition engine. |
|
| Solution: | |
| Resolve this problem by replacing referenced grammars whose language property does not match the default language value of the application with grammars whose language value is identical to the application's default language. | |
|
|
|
| Problem: | |
|
Using Conversational Grammar Builder, when I click Parse to parse my grammar, I get the following error:
My conversational grammar references a keyword (.grxml) grammar.
This error indicates a mismatch between the language of the speech recognition engine that Conversational Grammar Builder is using and the language of a grammar that Conversational Grammar Builder is attempting to load. For example, if the conversational grammar in your English (United States) application references a French (Canada) keyword grammar, parsing the conversational grammars generates this error because Conversational Grammar Builder is attempting to load and parse a French keyword grammar using an English speech recognition engine. |
|
| Solution: | |
| Resolve this problem by replacing referenced grammars whose language property does not match the default language value of the application with grammars whose language value is identical to the application's default language. | |
|
|
|
| Problem: | |
| When I am using a Concept Recognition grammar that employs several Concepts under a common Concept Answer I notice that the resultant grammar requires more CPU to run effectively. | |
| Explanation: | |
| This is to be expected and is a consequence of the fact that, typically, these grammars have an order of magnitude more training sentences than the equivalent Keyword Answers, and therefore the resultant Statistical Language Model generated is more complex. | |
|
|
|
| Problem: | |
| After renaming a nested custom activity on the workflow canvas, the next instance of the nested custom activity fails to add. | |
| Solution: | |
Resolve this problem by doing one of the following:
|
|
|
|
|
| Problem: | |
When an application makes a speak request that specifies an audio source that cannot be fetched (regardless of the reason), the alternate text will not be rendered and the speak request will complete with an error having the following inner exception:
For example, the following SSML contains a localhost reference to an audio source file named "MyPrompt.wav":
<ssml:audio src="http://localhost/MyPrompt.wav">In this example, if the audio source file does not exist, the alternate text "My alternative text" will not be played. Furthermore, when an audio source reference cannot be fetched, the event log will contain an error message similar to the following message: |
|
| Solution: | |
| To resolve this problem, do not rely on alternate text for audio source references. If possible use a prompt database instead. If you need to use an audio source reference in your prompt, you can simulate the alternate text feature by replaying the prompt with the desired alternative text when a speak error is encountered. | |
|
|
|
| Problem: | |
| Section 6.3.4 of the VoiceXML 2.0 specification (http://www.w3.org/TR/voicexml20/#dml6.3.4) states that there are two types of barge-in available: speech or hotword. The hotword type is not supported by the Speech Server VoiceXML interpreter. If it is used in a VXML application, it will be ignored and no error will be thrown. | |
| Solution: | |
Resolve this problem by removing instances of the hotword type in your application.
Note Removing instances of the hotword type will cause barge-in behavior to default to the speech type of behavior, which is described in the VoiceXML specification as follows: "The prompt will be stopped as soon as speech or DTMF input is detected. The prompt is stopped irrespective of whether or not the input matches a grammar and irrespective of which grammars are active." |
|
|
|
|
| Problem: | |
| When I select Hang Up in the Voice Response Debugging Window while a prompt is playing, the CallDisconnectedException does not appear to be handled properly by the debugging session. | |
| Solution: | |
| Resolve this problem by disabling the Enable Just My Code (Managed only) debugging option in Visual Studio. To find this option, in Visual Studio click Tools, click Options, and then click Debugging. | |
|
|
|
| Problem: | |
| In the topic "64-Bit Issues" there is a procedure under the heading "Debugging Managed Code Voice Response Applications on Windows Vista." The documentation states that the Visual Studio 2005 Integrated Development Environment (IDE) can time out when multiple instances of the W3wp.exe process attempt to run without having correct permissions. Following the procedure is supposed to prevent the problem. However, when I try to perform step 5, I cannot find a %WINDIR%\Temp\SpeechWorkingFiles\Runtime directory. | |
| Solution: | |
| The directory in question is not created until the problem that this procedure was intended to prevent actually occurs. Wait for the problem to occur, and then follow the procedure to prevent it from happening again. | |
|
|
|
| Problem: | |
| Debugging an application over Remote Desktop works, but the use of direct audio input via the microphone is not supported. | |
| Solution: | |
| Resolve this problem by only using text or UI element input when debugging applications over Remote Desktop. | |
|
|
|
| Problem: | |
| While debugging an application on a computer running a 64-Bit operating system, when a call is made with the debugger, I receive the following C runtime exception: | |
| Solution: | |
| Resolve this problem by clicking OK and continuing. The application debugging should continue without any further issues. | |
|
|
|
| Problem: | |
| When I debug an application, I get an error message similar to the following: This problem is caused by the order in which Visual Studio and IIS are installed relative to one another, and occurs only when IIS is installed after installing Visual Studio. | |
| Solution: | |
Reregister ASP.NET. To re-register ASP.NET:
|
|
|
|
|
| Problem: | |
| When I debug my application, execution breaks after a wave file is submitted and Speech Server throws the following exception: This also happens when I try to use live audio for debugging instead of a wave file. | |
| Solution: | |
|
To resolve this problem, temporarily disable calls to the EnableRetainAudio() method on the SpeechRecognizer object while debugging your application.
When submitting either live audio or a wave file, the Speech Debugger internally restarts the recognition process in order to confirm the text of the recognition. Because the Speech Server API does not allow rerecording to the file path specified in the call to EnableRetainAudio(), it throws the above exception. |
|
|
|
|
© 2007 Microsoft Corporation. All rights reserved.