I’ve released two new PERL modules:

Speech::Swift – a PERL interface to the Cepstral text-to-speech engine, Swift.

and

Speech::Swift::Simple – a simplified interface to Speech::Swift

The libswift shared library is required to support this code, which is included with every voice downloaded from Cepstral.

The reason for two releases, is that the Speech::Swift module exports all (well, almost all) the underlying functions of the libswift.so library, while Speech::Swift::Simple has a simplified interface to generate speech in a just a few function calls.

For example:

#!/usr/bin/perl

use Speech::Swift::Simple;

#
# create a new Speech::Swift::Simple with one channel audio, and 16bit encoding.
#
my $s = new Speech::Swift::Simple(
         channels => 1,
         encoding => Speech::Swift::AUDIO_ENCODING_PCM16
);

#
# set the voice to use by name
#
$s->set_voice("Allison");

#
# synthesize the text, and return it as a Speech::Swift::Simple::Wav object
#
my $wav = $s->generate("My name is allison");

#
# write the wav object to a file.
#
$wav->write("test.wav");

Or use the Speech::Swift library directly, for a more low-level interface.

The audio output is always as a WAV file; you can use one of the many audio modules available from CPAN, like Audio::GSM or Audio::MPEG, to re-encode the audio as needed.

Both PERL modules are available for download from CPAN now.