ScanSoft TTS 1

[Archangel58]

Does anyone know if there are configuration settings (other than "Voice Speed" in Control Panel) that would improve the quality of the ScanSoft TTS output? We are using TTS version 2.0 and Voicemail Pro 4.1.40 with IMS on a P4 3.2GHz server with 1.5GB of RAM.

Any help would be tremendously appreciated.

 

[intrigrant]

Text to Speech SAPI Controls


Windows TTS engines use Microsoft's SAPI (Speech Application Program Interface). This includes the use of XML tags in the text to change how the text is spoken. For example, in the text This is the <volume level="90">text</volume> to speak the items within < > brackets are XML tags used to change how the speech is spoken.

The following are samples of the SAPI XML controls supported by Voicemail Pro TTS. Further information on SAPI 5 can be obtained from Microsoft support websites.

• Volume - Change the speech volume.

• Rate - Change the speech rate.

• Pitch - Change the speech pitch.

• Emph - Add emphasis to words.

• Spell - Spell out words and numbers literally.

• Silence - Add a period of silence.

• Partofsp - Change the usage of words.



Entering XML Tags
XML tags can be used in two ways, either nested or empty.

• Nest Tags: Example = <volume level="90">text</volume>
Nested tags consist of:

• An opening XML tag, in the example above <volume level="90">.

• The text to which the opening tag command should be applied.

• A closing XML tag. The closing tag use the same command as the opening tag, prefixed with / and no other settings. In the example above this is </volume>.

• Empty Tags: Example = <volume level="90"/>All following text

An empty tag is not nested, its command and settings apply to all the following text. Empty tags are recognizable by the / before the tags closing >.



Example SAPI XML Tags


Volume
Controls the volume of the speech. This tag can be nested or empty.

• Attributes:

• level=
Supports values between 0 and 100, being percentages of the system's set volume.

• Examples:

• <volume level="50"/>Speak allow following text at level 50.

• <volume level="50">Speak this text at level 50</volume> and this as normal.


Rate
Controls the speed at which the text is spoken. This tag can be empty or nested.

• Attributes
Only one attribute may be applied within a tag.

• absspeed=
Sets the absolute speed for the speech in a range between -10 and 10 with 0 being normal speech.

• speed=
Sets a speed change that is added to the current speed.

• Examples:

• <rate absspeed="5">Speak this text at rate 5</rate> and this text as normal.

• <rate absspeed="5"/>Speak all following text at rate 5.

• <rate speed="-5"/>Drop the current speech speed by 5.


Pitch
Controls the pitch at which the text is spoken. This tag can be empty or nested.

• Attributes
Only one attribute may be applied within a tag.

• absmiddle=
Sets the absolute pitch for the speech in a range between -10 and 10 with 0 being normal speech.

• middle=
Sets a pitch change that is added to the current speed.

• Examples:

• <rate absmiddle="5">Speak this text at pitch 5</rate> and this text as normal.

• <rate absmiddle="5"/>Speak all following text at pitch 5.

• <rate middle="-5"/>Drop the current speech pitch by 5.


Emph
Applies emphasis to a word or section of text. This tag must be nested.

• Attributes:
This tag has no attributes.

• Example:

• Say <emph>boo</emph>.


Spell
Spell forces the engine to speak any nested text literally rather than applying any speech rules. This is useful for numbers where rather that than saying 3001 as three thousand and one for 3001, the speech required is three zero zero one. This tag must be nested.

• Attributes
This tag has no attributes.

• Example:

• The telephone number is <spell>555 3468</spell>.


Silence
Inserts a period of silence. This tag must be empty.

• Attributes

• msec=
Sets the duration in milliseconds.

• Example:

• A short silence <silence msec="500"/> of half a second.


Partofsp
Forces the pronunciation of a word according to it usage if not correctly determined by the TTS speech engine or to override the engine. This tag must be nested.

• Attributes:

• part=
Takes a value from noun, verb, modifier, function or interjection.

• Example:

• To <partofsp part="verb">record</partofsp> that <partofsp part="noun">record</partofsp> press 1.

 

[Archangel58]

Thanks intrigrant,

As usual your input is "in"valuable and "in"sightful.