DrunkenBlog

Domo Arigato, Mr. Roboto

Awhile ago, I posted this goofy MP3 Jesper had made of my goofier interview, spoken via the built-in voices on the Mac. Jesper used a perl script, along with the 'say' command. The say command is...

• A CLI tool included with Mac OS X which pumps your text through the old speech synthesis engine that's also included, which can be sent to your speakers or to an .AIFF file.

• Is a fantastic way to play a prank on someone via SSH, along with a bit of AppleScript to set the volume (Not going to give it to you, but think 'tell application "System Events" to set volume')

• The largest improvement to speech on the Mac since it was first introduced with the AV Macs almost 13 years ago that I'm aware of. I'm really not kidding, although it was ported to OS X.

  It's one of those technologies where I don't know if it wasn't really used because it was a ahead of its time, the implementation sucks, or the whole ideal sucks in general, but it's one I keep in mind during every hype cycle.

Anyways, some of you thought it was kind of a clever idea, I certainly did, and Jesper threw together the script he calls "Interview to Speech" under the BSD license with some basic instructions here...

[ Interview to Speech ]

Do remember this is really just a w00t hack Jesper put together, and it's not for everyone, even those that understand the prerequisites:

You'll need to have a clue how to install Perl modules and the "LAME" software to be able to configure your Mac to use this script. A general sense of how to use *nix-style commands, pipes and Perl are preferable. If you don't know how to do these things, try to get someone that does know these things to help you, or turn back at once. This is more of a *nix hack than a Mac application - don't expect the latter.

Since the say command pipes out a big AIFF file, LAME is used to stunt it down to an .MP3, and Perl is used for parsing and decoding entities and such, and an app named mp3cat is used which is why you have to install some perl modules via CPAN.

You're on your own with Perl on the Mac, because I've tried to avoid using it on OS X after some bad experiences... but for LAME, while (as mentioned in the script you can get it via Fink or Darwinports), that may seem overkill if you just need one thing, and there are a whole slew of compiled binaries of LAME out there (and it has few dependencies, so building from source isn't a nightmare).

The other thing you need to deal with is, well, letting the script know which parts of the text should be spoken with which voice, which the script notes in the comments and just involve an H3 tag for the interviewee. Pretty simple, so maybe someone will do something suitably goofy with it.

 Posted by drunkenbatman
    July 13, 2005, at 11:47 AM