Phoneme mode input to speech synthesizer on OS X

Question

Working on the built-in speech synthesis of OS X, which can be tested quite easily directly from the command line by using the say command.

According to the documentation, the following string should provide improved pronunciation:

My name is [[inpt PHON]] AY1yIY2SAX [[inpt TEXT]].

This works from the command line with the voice "Alex":

say -v Alex 'My name is [[inpt PHON]] AY1yIY2SAX [[inpt TEXT]].'

However, if you try it with another voice, such as "Samantha", it does not seem to work:

say -v Samantha 'My name is [[inpt PHON]] AY1yIY2SAX [[inpt TEXT]].'

Samantha (or "Sam" to her friends) ignores the command in the double square bracket and just spells out the phonemes as latin letters. Not what I wanted.

Note that Sam does recognise other commands in the string:

say -v Sam 'The code is [[nmbr LTRL]] 1989[[nmbr NORM]].'

As opposed to:

say -v Sam 'I was born in 1989.'

So the question is: Is there a way to get Sam and other stubborn voices to change the pronunciation or to accept phonemic input?

I have tested this on OS X 10.11.6.

Answer 1

Apparently, phoneme input doesn’t work with the Nuance voices. It’s a shame.