meSpeak configuring female voice

Question

I was following the following post to configure the female voice to meSpeak JavaScript variation of eSpeak TTS framework.

Link

But I am unable to get it to work with the female voice. My en-us-f.json file look as the following.

    {
"voice_id":"en/en-us-f",
"dict_id":"en_dict",
"dict": <same as the dict base 64 encoded default>
"voice":"name english-us-f\nlanguage en-us 2\nlanguage en-r\nlanguage en 3\ngender female\nphonemes en-us\ndictrules 3 6\noption reduce_t 1\nstressLength 140 120 190 170 0 0 255 290\nstressAmp  17 16  19 19  19 19  21 19\nreplace 03 I  i\nreplace 03 I2 i",
"voice_encoding": "text"
}

Please help me to configure this properly. What am I doing wrong?

Answer 1

Here is the code used for the en-us voice in http://www.masswerk.at/eliza – it's made to sound a bit like the cliche of an about 50-ish female psychologist.

{
"voice_id":"eliza/en/en-us",
"dict_id":"en_dict",
"dict":"AAQAAECpAAANRUE(... snipp ...)",
"voice":"name eliza-english-us\nlanguage en-us 2\nlanguage en-r\nlanguage en 3\ngender female 50\nphonemes en-us\ndictrules 3 6\noption reduce_t 1\npitch 90 138.\necho 40 20\nintonation 3\nflutter 1\nroughness 1\nvoicing 80\nconsonants 103 120\ntone 600 160  1200 145  2000 120\nstressLength 145 125 190 170 0 0 260 290\nstressAmp  17 16  19 19  19 19  21 19\nstressAdd 0 0 -10 -10 0 0 10 40\nbreath 0 2 3 3 3 3 0 5\nbreathw 150 150 200 200 400 400\nformant 0 105  80 150\nformant 1 110  80 160\nformant 2 110  70 150\nformant 3 110  70 150\nformant 4 115  80 200\nformant 5 115  80 100\nformant 6 110  70 150\nformant 7 110  70 100\nformant 8 110  70 150\n",
"voice_encoding":"text"
}

For the full file see: http://www.masswerk.at/eliza/voices/eliza-en-us.json

Make sure to have at least meSpeak.js version 1.3.1 (see heading comment in mespeak.js; current version is 1.5, update recommended).

Editing Best Practice: The internal eSpeak core will just fail on a malformed voice-syntax. Try again with some of the more complex lines deleted and work from a working voice description forward to a more sophisticated one.

e.g. you could start with:

"voice": "name english-us-f\nlanguage en-us 2\nlanguage en-r\nlanguage en 3\ngender female\nphonemes en-us\ndictrules 3 6\noption reduce_t 1"

and start adding stress and replacements options to this.