Change the style of the default voice of Speak function?

Questions
, , ,
1

Dear Community,

Speak is a nice and easy to use function to tell written text. The default voice is a pleasant female voice. Is there a way to change it to something else? I tried to figure it out from the help, but failed.

Tx for the help in advance,
best regards

Andras

2

Use SpeechSynthesize that give access to several voices available on your system. You can query for available voices and their properties using VoiceStyleData[].

Once you have the audio, you can use AudioPlay to play it, and AudioPause or AudioStop to pause/stop it.

3

Works perfectly :slight_smile: Thank you very much.
Best regards,
Andras

4

Hi Shadi

VoiceStyleData[] show 52 voices in both local docs and online docs: VoiceStyleData: Text to speech voices—Wolfram Documentation

I only get 5 voices when running VoiceStyleData[] in my notebook:
{“AlanWBlack”, “KevinLenzo”, “KevinLenzo16”, “RMS”, “SLT”}
same for $VoiceStyles

Is it possible to change the voice for Speak[]?

How is the Speak[] voice generated?

Thanks

5

The list of voices returned by VoiceStyleData and $VoiceStyles should be the same on the same machine. They vary depending on your platform. We use some generic voices like “AlanWBlack” that are available on all platforms, then on Windows and Mac, we do use operating system voices that are platform specific.

Speak respects system settings. For example, on mac, you can change the voice in System Preferences > Accessibility > Speech. Default voice in SpeechSynthesize does not currently respect system settings.

There are obviously some consistency updates and additional utilities and voice properties that we can introduce here.

6

Hi Shadi,

I’m using Mathematica 12 on linux (ubuntu). Haven’t tested the accessibility voice, but I see there is at least a menu for it.

Is there anyway to load the missing voices?

I’m trying to build a simple cloud “spelling bee” program for a 2nd grader. Enter list of words, spell the entire word, spell the letters of each words. Trying to figure out what are the best options for (a) spelling out a word’s letters, and (b) speaking out an entire word.

SpeechSynthesize[] is more robust, but inherits the interface. Speak[] is faster but lacks features and the voice sounds a bit too robotic.

Is it possible to incorporate Speak[] or SpeechSynthesize[] in a clouddeploy form? a local RPI3 is also an option but I’d prefer a cloud option.

Any suggestions/recommendations?

Thanks.

b