Package: festival Version: 2.1~release-1 Festival 2.1 comes with new integrated support for using clustergen voices via the clustergen module. The upstream original tar automatically includes this module when building on a Debian system. Festival 2.1~release-1 build system does not build this. The custom Debian build scripts carry over from earlier versions of festival and inclusion of this module was apparently missed when Debian moved to festival 2.1.
I attach a patch to the existing Debian build patch to include this module. It should be noted that clustergen voices have a more modern design than traditional diphone voices and are widely reported to have more "human-like" qualities for similar size voice files. with very best regards, Peter
diff -Naur a/debian/patches/config_config.diff b/debian/patches/config_config.diff --- a/debian/patches/config_config.diff 2011-08-18 16:42:59.000000000 +1000 +++ b/debian/patches/config_config.diff 2011-08-26 10:13:24.000000000 +1000 @@ -63,7 +63,7 @@ +# ALSO_INCLUDE += UniSyn_phonology UniSyn_selection +## Cluster unit selection code as described in "Building Voices in +## Festival", again experimental and suitable for research purposes only. -+ALSO_INCLUDE += clunits hts_engine MultiSyn ++ALSO_INCLUDE += clunits clustergen hts_engine MultiSyn + +## Old diphone code that will be delete, left in only for some +## compatibility
