Change voice mid text about voices

[mklement0]

@2V3EvG4LMJFdRe

You need to provide the text via the pipeline / stdin:

echo '[[voice Alex]] I am Alex. [[voice Samantha]] I am Samantha. [[voice default]] Back to default.' | say-multivoice

The original script was bare-bones, it was just a quick proof of concept. I've updated it to be a little friendlier, with command-line help and error checking - please see above.

[2V3EvG4LMJFdRe]

Thanks a lot! I can finally do what I requested, so I'm gonna close the issue.

[mklement0]

That's an interesting idea, but note that voices just defers to say, which uses the speech synthesizer API - and it would have to be that API that supports commands to change the speaking voice.

I didn't actually know about "markup" such as [[slnc 1000]], which Apple calls embeddable speech commands.

I couldn't find current documentation, but here's an archived version that suggests that at least historically changing the voice wasn't supported: https://developer.apple.com/library/archive/documentation/UserExperience/Conceptual/SpeechSynthesisProgrammingGuide/FineTuning/FineTuning.html#//apple_ref/doc/uid/TP40004365-CH5-SW12

You could ask a question at https://apple.stackexchange.com/ to see if someone has a definitive answer for the current macOS version.

Building such functionality on top of the API is conceivable, but I think is outside the mandate of voices.

If you wish to explore this further yourself, here's a quick PowerShell prototype, which wraps the say utility:

# Function definition
function say-multivoice {
    $voiceArg = @()
    $Input -split '(\[\[voice .+?\]\])' -ne '' | foreach {
        if ($_ -match '\[\[voice (.+?)\]\]') {
            $voice = $Matches[1]
            if ($voice -eq 'default') { $voiceArg = @() }
            else                      { $voiceArg = '-v', $voice }
        } else {
            $_ | say $voiceArg
        }

    }
}

Once the above function is defined (e.g. via your $PROFILE file), you could use commands such as the following:

'[[voice Alex]] I am Alex. [[voice Samantha]] I am Samantha. [[voice default]] Back to default.' | say-multivoice

[2V3EvG4LMJFdRe]

Thanks a lot for the thoughtful response, it goes very in depth at what I've found is a rather obscure operation. Since I'm not a programmer, some of what you write goes over my head. From what I gather, I have to install PowerShell, and then create a function that calls for the voices to change on a $PROFILE file, which would link it for use system-wide. I've installed PowerShell with brew, but now I'm not sure where the $PROFILE file is located to put the above in.

[mklement0]

On second thought, I suggest creating an executable shell script, so that you can invoke it from any shell, without having to start PowerShell explicitly:

Assuming you have administrative privileges on your machine, open a terminal and run the following commands; you'll be prompted for your login password once:

• Create script file /usr/local/bin/say-multivoice

  • sudo touch /usr/local/bin/say-multivoice

• Make it executable:

  • sudo chmod +x /usr/local/bin/say-multivoice

• Open the script file for interactive editing:

  • sudo nano /usr/local/bin/say-multivoice

• Once the interactive editor (nano) is up, paste the following text:

#!/usr/bin/env pwsh -noprofile

# See https://github.com/mklement0/voices/issues/6#issuecomment-666714554

if ($args[0] -in '-?', '-h', '--help') {
@"
Wrapper for the `say` utility that supports text with embedded commands to
switch the speaking voice.

Usage: 
  <text> | say-multivoice
  say-multivoice < file              # not in PowerShell
  Get-Content file | say-multiveoice # PowerShell

Example:
   echo '[[voice Alex]] I am Alex. [[voice Samantha]] I am Samantha. [[voice default]] Back to default.' | say-multivoice
"@
  exit 0
}
elseif ($args.Count) {
  [Console]::Error.WriteLine(@"
ERROR: Unexpected argument(s) specified. Please provide the text to speak via the pipeline (stdin). Use -? for help.
"@)
  exit 1
}
elseif (-not $MyInvocation.ExpectingInput) {
  [Console]::Error.WriteLine(@"
ERROR: Missing input. Please provide the text to speak via the pipeline (stdin). Use -? for help.
"@)
  exit 1
}

$voiceArg = @()
$Input -split '(\[\[voice .+?\]\])' -ne '' | ForEach-Object {
    if ($_ -match '\[\[voice (.+?)\]\]') {
        $voice = $Matches[1]
        if ($voice -eq 'default') { $voiceArg = @() }
        else                      { $voiceArg = '-v', $voice }
    } else {
        $_ | say $voiceArg
    }
}

exit $LASTEXITCODE

• Press Control-X, then y, then Return to save the file.

After that, you should be able to call say-multivoice from any terminal; e.g.:

echo '[[voice Alex]] I am Alex. [[voice Samantha]] I am Samantha. [[voice default]] Back to default.' | say-multivoice

However, note that all you can do is pipe text to it, none of say's other options are supported by this simple wrapper.

[2V3EvG4LMJFdRe]

Having followed your instructions, I'm not sure how to put it to work. say-multivoice is indeed recognized as a terminal command, but I'm not sure what the syntax of it would be so that it can speak text out loud. For example, entering the line say-multivoice '[[voice Alex]] I am Alex. [[voice Samantha]] I am Samantha. [[voice default]] Back to default.' doesn't prompt any speech.