Entering Transcriptions

Article
07/05/2006

Use the Recording and Extraction panes in Speech Prompt Editor to enter transcriptions for prompt recordings, to create word and phrase components that are built into prompt sentences, and to record prompts. Transcriptions are the texts that users read to create recorded prompts.

Use the Options dialog box for Speech Prompt Editor to choose options such as recording format and frequency, and default column display.

To open Speech Prompt Editor

Open a prompt project.
In Solution Explorer, double-click a prompt database.

Using the Keyboard to Add Transcriptions

The prompts database contains transcriptions. Speech Prompt Editor records transcriptions as .wav files that are also stored in the prompts database, along with tracking and status data. Begin building a prompt database by adding transcriptions. To add transcriptions to the database, enter them manually using the keyboard, or import them.

Note Speech Prompt Editor consists of two windows. The Transcription pane is the top window, and the Extraction pane appears below the Transcription pane.

To enter transcriptions manually

Open a prompt database.
In the Transcription pane, click once to select a row, then click again in the Transcription cell to open an edit box.
Type a transcription, for example, "Welcome to Paris", and then press ENTER.

After you press ENTER, icons appear in the Has Wave and Has Alignments columns, indicating that the transcription is not yet associated with a .wav file, and that there are no word alignments. When the user records a transcription, Speech Prompt Editor creates alignments, as described in Recording Transcriptions.

Customizing Display of Properties and Columns

Customize the display in the Recording and Extraction panes by selecting which columns to display and sorting rows by the values in a particular column.

Selecting Columns for Display

In the Recording and Extraction panes, right-click a column header to display a list of columns. Select check boxes next to the column names to display them.

Sorting Rows

In the Recording and Extraction panes, click a column header to sort rows by the contents of the column. Click again to toggle between ascending and descending sort orders. Click a third time to return to the original sort order.

Displaying Field Values in the Properties Window

For both the Recording and Extraction panes, the values of the selected row also appear in the Properties window. This feature enables the user to select one row and scroll away to view other rows while still displaying values from the original row. Users can choose not to display a column, because that column's values are still visible in the Properties window.

Creating Extractions

An extraction is a segment of a prompt that can combine dynamically with other extractions at run time to create a prompt. For example, the three extractions "medium," "latte," and "You ordered a" can combine to create the prompt "You ordered a medium latte." The built version of the prompt database, the .prompts file, contains only extractions. Parts of a sound recording that are not marked as an extraction are not available for the application.

Using extractions avoids the expense of recording all the variations of repetitive prompts such as "You ordered a medium latte ... You ordered a double espresso ... You ordered a small coffee." To create extractions, place brackets ([]) around the selection in the Transcription pane.

Note Some punctuation does not appear in the extractions, even though it appears in the display text. If the punctuation does not appear in the transcription or extraction, using it in prompt text results in the text-to-speech (TTS) engine playing the prompt. For instance, if "A - B" is entered in the Transcription field, it appears as "A B", even though the Display Text field shows "A - B", and it appears in the Extraction pane as "A B", with no hyphen. If "A - B" is used in prompt text, the TTS engine renders it as "A B".

It is also possible to identify each extraction with a user-defined Extraction ID. This facilitates automated maintenance of prompt databases, as well as making it easier to keep track of extractions. In addition, Extraction IDs can appear in PEML markup, to specify the ID of a particular prompt to play.

To create extractions

Open a prompt database.
In the Transcription pane, select a row containing a transcription, then click the Transcription cell.
In the Transcription column, enter opening and closing square brackets ([]) around a word or phrase.
Press ENTER.

To add an Extraction ID to an extraction

In a transcription in the Transcription pane, add an ID in angle brackets () inside the square brackets marking off an extraction.
```
Hello, [Scott <scottname>]
```
- or -
In the Extraction ID column of the Extraction pane, type an ID.

Creating Tags

Use tags to identify varying recordings of the same text. For example, the word "medium" might be recorded both in a question and in a statement. Marking each recording with a unique tag allows the application to select the correct recording at run time. Enter tags in the Recording, Extraction, or Properties window. Use the withtag element in Prompt Engine Markup Language to specify tagged prompt database entries.

Speech Prompt Editor displays tags in the following three locations.

Transcription column in the Transcription pane
Tag column in the Extraction pane
Extraction property in the Properties window

To enter tags in the Transcription pane

Open a prompt database.
In the Transcription pane, select a row containing an extraction, then click the Transcription cell.
In the Transcription cell, within the brackets enclosing the extraction, enter tag text with curly braces ({}) surrounding the tag text.
Press ENTER.

To enter tags in the Extraction pane

Open a prompt database.
In the Extraction pane, select a row containing an extraction, then click the Tag cell.
Enter the tag text in the Tag cell.
Press ENTER.

To enter tags in the Properties window

Open a prompt database.
In the Extraction pane, select a row containing an extraction.
In the Properties window, select the Tag property and enter tag text.
Press ENTER.

Creating Multiple Tags

Multiple tags can be associated with a single extraction. Use semicolons as a separator when entering multiple tags.

The following transcription markup creates both a name tag and a male tag. When used in an application, a prompt can request either the name tag, the male tag, or both tags to identify this particular extraction.

  Hello [Scott {name;male}]

Note If a prompt requests but cannot find an extraction that contains all (or more) of the requested tags, the prompt engine searches only for an extraction that matches the transcription, and not the extraction with the most matching tags.

Entering Valid Characters

Valid characters in the Transcription pane include those in the following list.

any alphanumeric character in the ranges a-z, A-Z and 0-9
angle, square and curly brackets (, [], {})
any character contained in a word, for example the words don't and he#%llo are both valid

Formatting Data for Speech

Use text normalization functions to control the way in which data is spoken in prompts. Without a text normalization function, the TTS engine speaks the phone number (555) 555 5555 as "open paren five hundred fifty five close paren five million five hundred fifty five thousand five hundred fifty five." Adding a text normalization function allows the prompt engine to use recorded prompts and speak this phone number as "area code five five five (pause) five five five (pause) five five five five." Text normalization functions are useful for formatting a variety of data.

Description	Example
Credit card numbers	1234 5678 9012 3456
Dates	04/13/85
Phone numbers	(111) 222-3333
Abbreviations	Mr. (Mister), Dr. (Doctor or Drive), St. (Street or Saint), Oz (ounce), Lb (pound)
Addresses	1 Microsoft Way Redmond, WA 98025

Add text normalization functions to transcription files in the prompt project. Use the peml:rule element in Prompt Engine XML to specify which text normalization function to apply to the prompt text. See the following example, where the rule element references a prompt function called phone_number to speak the phone number 01-1-425-555-0100.

<peml:prompt_output>

<peml:rule name="phone_number">01-1-425-555-0100</peml:rule>

</peml:prompt_output>

To add a normalization script file to a prompt project

Open a prompt project.
In Solution Explorer, right-click the project, select Add, then select Add New Item.
In the Add New Item dialog box, in the Templates pane, double-click either the VBScript File icon or the JScript File icon.

Partager via