View on GitHub

SALB

Frontend for speech synthesis using HTS voice models.

Download this project as a .zip file Download this project as a tar.gz file

SALB

The SALB system is a frontend framework for speech synthesis using HMM based voice models built by HTS. See our Releases here and the github project page here. Build instructions, license information and a generic readme are contained in the package. Additional documentation can be found in the project wiki.

Facts & features:

The SALB framework

We also provide a free (Austrian) German voice model of a male, professional speaker, including number conversion, a pronunciation lexicon with 14k entries and a set of Letter-To-Sound rules. Listen to some samples here: Sample 1, Sample 2, Sample 3

F.A.Q.

How can I create a new voice model?

An easy way is to use the HTS demo packages. These produce htsvoice files that can be used with the SALB system. For more information, take a look at the project wiki.

Why not use flite+hts_engine instead?

The SALB system is another layer on top of flite and hts_engine (see Architecture). You might want this additional layer if:

*Of course this is also possible in flite but a bit trickier. If you would really like to do that, take a look our description on adding a new language to flite here.

How can I contribute?

We would love to include language bindings (e.g. Python), further interfaces and more languages. Of course, bug fixes, improvements etc. are also always welcome.