Scripting: SIExtractPhonemes

SIExtractPhonemes

Introduced

v9.0 (2011)

Description

Extracts phonemes from an audio file using a text to guide the recognition

Scripting Syntax

oReturn = SIExtractPhonemes( Language, FileName, GuideText );

Return Value

Returns a multi-dimentional array of results. Each column will contain

1) The Unicode IPA code of the phoneme

2) The Start Time of the phoneme, in seconds

3) The End Time of the phoneme, in seconds

4) A human-readable hint of phoneme, useful for debugging

Parameters

Parameter	Type	Description
Language	String	Language to use. "en" for English, or "jp" for Japanese
FileName	String	Path to an audio file, for example a .wav or an .avi file. Mono and 16000k is recommend. Same audio formats as Audio in Softimage; MP3 not supported.
GuideText	String	Text that match the spoken words in the audio file. Avoid any unusual punctuations or strange characters that may throw off the engine

Examples

JScript Example

//Returns a VBArray object, which require using the .getItem method in Jscript because they are not 
//native objects there.  However, they work transparently like multi-dimentionnal arrays in VBscript and Python
var res = SIExtractPhonemes ("en", "s:\\here_be_dragons.wav", "Here be dragons, the map says");
arraysize = res.ubound( 1 );
logmessage( arraysize + "phenemes found" );
for (  i=0; i < arraysize; i++)
{
logmessage( "IPA#" + res.getItem(i,0) 
+ "(" + res.getItem(i,3) + ") "
+ "Start Time(sec) " + res.getItem(i,1) 
+ "End Time(sec) " + res.getItem(i,2) );
}