How can I enumerate available voices and languages using espeak API?

Question

I am using espeak API from C++ to do some simple text to speech synthesis from my embedded app. Currently, I have copied this line from the basic example on how to get started:

espeak_SetVoiceByName("default"); 

This seems to work fine, however I know that espeak comes with several voices in several different languages. How can I enumerate those and later select them using espeak API?

Answer 1

If you have installed the package you might launch espeak directly from the command line, and in this case

espeak --voices

will enumerate all available voices. Just an excerpt from my machine:

Pty Language Age/Gender VoiceName          File          Other Languages
 5  af             M  afrikaans            other/af      
 5  an             M  aragonese            europe/an     
 5  bg             -  bulgarian            europe/bg     
 5  bs             M  bosnian              europe/bs     
 5  ca             M  catalan              europe/ca     
 5  cs             M  czech                europe/cs

I've found out that both VoiceName and Language can be used in the espeak_SetVoiceByName API

Answer 2

The documentation for espeak API is the headerfile itself. You can find it here.

To enumerate existing voices, you can use something like this:

const espeak_VOICE **list=espeak_ListVoices(0);
espeak_VOICE *voice=0;
for(;*list!=0;++list){
    voice=*list;
    if(0!=voice){
        //Look at voice parameters such has voice->name here
    }
}

Later when you have found a voice you want to use, you can set it with this:

if(0!=voice){
    espeak_SetVoiceByProperties(voice);
}

The espeak_VOICE struct is defined like this:

typedef struct {
    const char *name;      // a given name for this voice. UTF8 string.
    const char *languages;       // list of pairs of (byte) priority + (string) language (and dialect qualifier)
    const char *identifier;      // the filename for this voice within espeak-data/voices
    unsigned char gender;  // 0=none 1=male, 2=female,
    unsigned char age;     // 0=not specified, or age in years
    unsigned char variant; // only used when passed as a parameter to espeak_SetVoiceByProperties
    unsigned char xx1;     // for internal use
    int score;       // for internal use
    void *spare;     // for internal use
} espeak_VOICE;

Answer 3

Use the espeak_SetVoiceByProperties function that's defined directly below the one you used.

#ifdef __cplusplus
extern "C"
#endif
ESPEAK_API espeak_ERROR espeak_SetVoiceByName(const char *name);
/* Searches for a voice with a matching "name" field.  Language is not considered.
   "name" is a UTF8 string.

   Return: EE_OK: operation achieved
           EE_BUFFER_FULL: the command can not be buffered;
             you may try after a while to call the function again.
       EE_INTERNAL_ERROR.
*/

#ifdef __cplusplus
extern "C"
#endif
ESPEAK_API espeak_ERROR espeak_SetVoiceByProperties(espeak_VOICE *voice_spec);
/* An espeak_VOICE structure is used to pass criteria to select a voice.  Any of the following
   fields may be set:

   name     NULL, or a voice name

   languages  NULL, or a single language string (with optional dialect), eg. "en-uk", or "en"

   gender   0=not specified, 1=male, 2=female

   age      0=not specified, or an age in years

   variant  After a list of candidates is produced, scored and sorted, "variant" is used to index
            that list and choose a voice.
            variant=0 takes the top voice (i.e. best match). variant=1 takes the next voice, etc
*/

The espeak_VOICE structure is defined and documented not far above it.

The espeak_ListVoices function, to enumerate voices as requested, is defined right above the functions I quoted.