Copyright (c) Hyperion Entertainment and contributors.

Difference between revisions of "Translator Library"

From AmigaOS Documentation Wiki
Jump to navigation Jump to search
 
Line 1: Line 1:
 
== Translator Library ==
 
== Translator Library ==
   
This chapter describes the translator library which, together with the narrator device, provides the Amiga's text-to-speech capability. To fully understand how speech is produced on the Amiga, you should also read [[Narrator_Device|Narrator Device]].
+
This article describes the translator library which, together with the narrator device, provides the Amiga's text-to-speech capability. To fully understand how speech is produced on the Amiga, you should also read [[Narrator_Device|Narrator Device]].
   
 
The translator library provides a single function, Translate(), that converts an English language string into a phonetic string. You may then pass this phonetic string to the narrator device which will say the string using the Amiga's audio hardware. The two subsystems may also be used individually. You don't have to use the narrator to say the phonetic strings; you could use them instead for phonetic analysis or some other special purpose.
 
The translator library provides a single function, Translate(), that converts an English language string into a phonetic string. You may then pass this phonetic string to the narrator device which will say the string using the Amiga's audio hardware. The two subsystems may also be used individually. You don't have to use the narrator to say the phonetic strings; you could use them instead for phonetic analysis or some other special purpose.
Line 63: Line 63:
 
The English language has many words that do not sound the same as they are spelled. The translator library has exception rules that it consults as the translation progresses. It also provides for common abbreviations such as Dr., Prof., LB., etc. Words that are not in the exception table are translated literally. This translation allows unrestricted English text as input, and uses over four hundred and fifty context sensitive rules. It automatically accents content words, and leaves function words (e.g. of, by, the, and at) unaccented. It is possible, however, that certain words will not translate well. You can improve the quality of translation by handling those words on your own.
 
The English language has many words that do not sound the same as they are spelled. The translator library has exception rules that it consults as the translation progresses. It also provides for common abbreviations such as Dr., Prof., LB., etc. Words that are not in the exception table are translated literally. This translation allows unrestricted English text as input, and uses over four hundred and fifty context sensitive rules. It automatically accents content words, and leaves function words (e.g. of, by, the, and at) unaccented. It is possible, however, that certain words will not translate well. You can improve the quality of translation by handling those words on your own.
   
The phoneme table that the narrator uses is listed in the [[Narrator_Device|Narrator Device]]. You will also find other important information about the Amiga's speech capability in the narrator chapter including a working example which shows how to use the translator library together with the narrator device.
+
The phoneme table that the narrator uses is listed in the [[Narrator_Device|Narrator Device]]. You will also find other important information about the Amiga's speech capability in the narrator article including a working example which shows how to use the translator library together with the narrator device.

Latest revision as of 21:26, 9 November 2015

Translator Library

This article describes the translator library which, together with the narrator device, provides the Amiga's text-to-speech capability. To fully understand how speech is produced on the Amiga, you should also read Narrator Device.

The translator library provides a single function, Translate(), that converts an English language string into a phonetic string. You may then pass this phonetic string to the narrator device which will say the string using the Amiga's audio hardware. The two subsystems may also be used individually. You don't have to use the narrator to say the phonetic strings; you could use them instead for phonetic analysis or some other special purpose.

Opening the Translator Library

To use the Translate() function, you must first open the translator library and obtain a pointer to its main interface:

struct Library *TranslatorBase;
struct TranslatorIFace *ITranslator;
 
TranslatorBase = IExec->OpenLibrary("translator.library", REVISION);
if(TranslatorBase != NULL)
{
  ITranslator = (struct TranslatorIFace*)IExec->GetInterface(TranslatorBase, "main", 1, NULL);
  if(ITranslator != NULL)
  {
    /* use translator here */
  }
}

Using the Translate Function

Once the library is open, you can call the translate function:

#define BUFLEN 500
 
STRPTR EnglStr;                 /* pointer to sample input string */
LONG EnglLen;                   /* input length */
UBYTE PhonBuffer[BUFLEN];       /* place to put the translation */
LONG rtnCode;                   /* return code from function */
 
EnglStr = "This is Amiga speaking.";    /* a test string */
EnglLen = strlen(EnglStr);
rtnCode = ITranslator->Translate(EnglStr, EnglLen, (STRPTR)PhonBuffer, BUFLEN);

The input string will be translated into its phonetic equivalent and can be used to feed the narrator device. If you receive a non-zero return code, you haven't provided enough output buffer space to hold the entire translation.

In this case, the Translate() function breaks the translation at the end of a word in the input stream and returns the position in the input stream at which the translation ended. You can use the output buffer, then call the Translate() function again, starting at this original ending position, to continue the translation where you left off. This method will sound smoothest if the ending position ends on sentence boundaries.

Translate() returns negative values
To get the proper character offset, you must use -(rtnCode) as the starting point for a new translation.

Closing the Translator Library

As with all other libraries of functions, if you have successfully opened the translator library for use, be sure to close it before your program exits. If the system needs memory resources, it can then expunge closed libraries to gain additional memory space:

struct Library *TranslatorBase;
struct TranslatorIFace *ITranslator;
 
IExec->DropInterface((struct Interface*)ITranslator);
IExec->CloseLibrary(TranslatorBase);

Additional Notes About Translate

The English language has many words that do not sound the same as they are spelled. The translator library has exception rules that it consults as the translation progresses. It also provides for common abbreviations such as Dr., Prof., LB., etc. Words that are not in the exception table are translated literally. This translation allows unrestricted English text as input, and uses over four hundred and fifty context sensitive rules. It automatically accents content words, and leaves function words (e.g. of, by, the, and at) unaccented. It is possible, however, that certain words will not translate well. You can improve the quality of translation by handling those words on your own.

The phoneme table that the narrator uses is listed in the Narrator Device. You will also find other important information about the Amiga's speech capability in the narrator article including a working example which shows how to use the translator library together with the narrator device.