Reco

Microsoft Speech SDK

The Microsoft.com Speech website Microsoft Speech SDK SAPI 5.1

Reco

Introduction

Reco is a speech tool that you can use to examine and test the speech process. You may speak using a microphone for either dictation or command and control. Alternatively, you may directly type a command and control order in an edit box and submit it to the speech recognition engine.

During the recognition process, events will be displayed in the top window as they occur. An event log is kept and to examine the details of each one, double-click the item. You may use this log to trace the sequence and number of events. You may also examine speech results. Double-click the text or phrase to display more detailed information about the text in the bottom window.

Options

The dialog box presents several options. By default, all options are initially on. The following options display detailed information suited for speech testing.

Create Recognition Context

Creates the recognition context for the selected engine. This must be off before selecting a different engine. Select this to activate the engine. You must select it before any speech is recognized.

Retain Reco Audio

Keeps the recorded sound of the actual spoken content in memory. Select this option and double-click the text in the display area of either the Events or Properties window to play back the spoken text. If it is not selected, the playback will be in a synthesized text-to-speech voice.

Activate Microphone

Controls the microphone's input. Select this option to enable the microphone to accept sound input. If it is not selected, the microphone will be blocked and no sound will be registered.

Load Dictation

Controls loading of dictation grammar. Select it to load the grammar for dictation and enable the speech recognition engine to process free-formed speech. You must load this grammar before selecting Activate Speech. If you do not select Load Dictation, the grammar will be unloaded.

Activate Dictation

Controls the activation status of dictation features. If you have loaded the dictation grammar (see Load Dictation), select this to allow speech attempts to be recognized as free-formed dictation.

Load Command and Control

Controls loading of command and control grammar. Select this to load the grammar for dictation and enable the speech recognition engine to process command-oriented speech. You must load this grammar before selecting Activate command and control. If you do not select Load Command and Control, the grammar will be unloaded.

Activate Command and Control

Controls the activation status of command and control features. If you have loaded the command and control grammar, (see Load Command and Control), select this to allow speech attempts to be recognized as commands according to the rules defined the in current grammar. By default, the Solitaire grammar is installed. However, you may change the grammar using the Command and Control->Load Grammar menu item.

Load Spelling

Enables or disables the letter-by-letter spelling feature. You must enable spelling before you select Activate Spelling. If Load Spelling is not selected, it will disable spelling.

Activate Spelling

Controls the activation status of the letter-by-letter spelling function. If you have loaded the spelling grammar, (see Load Spelling), select this to activate the spelling capability of the speech recognition engine. Using Activate Spelling, you can spell words in a letter-by-letter fashion while in dictation mode. For example, you can say, "Let's take the dogs on a w-a-l-k," and the engine will attempt to recognize the sentence as "Let's take the dogs on a walk."

Shared Recognizer

Selects the engine as a shared engine. Using Shared Recognizer, other speech-enabled applications running at the same time can use the microphone. To do this:
  1. In Control Panel, double-click the Speech icon.
  2. On the Speech Recognition tab, in the Speech Engine window, select an engine.

InProc

Selects an engine as an in-process (InProc) or non-shared engine. InProc restricts speech-enabled applications running at the same time from using the microphone. Because the engine will not be shared among other applications, using this option you can select different engines from the drop-down menu rather than using the Speech icon. Selecting an InProc engine will not change the engine used by shared applications or the one selected in the Speech Engine window.

Event Window

Displays events and any associated results. The name of the event is displayed inside brackets. Certain events such as recognition or a hypothesis, have text associated with them and that text is displayed immediately under the event. Double-click the text to hear it spoken. If Retain Reco Audio is currently selected, the playback will be in your voice, recorded when it was spoken; otherwise, the text-to-speech voice will be played back. You may change the synthesized voice using the TTS tab in Speech properties. Double-click the event to display additional information about it. The Results window will update to the new information.

Results Window

Displays details of the properties and values associated with the recognition result. This information includes the applicable rule, the text of the recognition, the parsing of information, and the individual elements of each word.

Alternates

Command and Control Edit Box

Enables you to directly enter a command and control phrase. If any text appears in the box, Submit activates it. Click Submit to enter a command, bypassing the microphone. In this way, the microphone does not need to be active or even present to test or watch the recognition process.

Submit

Submits any text in the command and control edit box, see Command and Control Edit Box.