IMS GERMAN FESTIVAL
(open source distribution)
version 1.4beta-os (based on Festival 2.5)
Institute of Natural Language Processing
University of Stuttgart, Germany
Copyright (c) 2019
All Rights Reserved.
This is information about the German Festival version developed at the
Institute of Natural Language Processing, University of Stuttgart by
Gregor Möhler, Antje Schweitzer, Mark Breitenbücher, Martin Barbisch,
and others (see ACKNOWLEDGEMENTS section in the manual for full list).
The IMS German Festival is a full German Text-to-Speech engine based
on the original Festival distribution by CSTR, University of
Edinburgh. Together with two non-IMS resources (the BOMP lexicon and
diphones from the MBROLA project, see below) it enables Festival to
This is version 1.4-os(beta), which has been tested only on Fedora Linux
so far. Theoretically, it should run fine on most Linux machines.
Basically the packages are distributed for research and educational
(i.e. non-commercial and non-military) purposes. The license terms can
be found in the file COPYING.ims_german_festival for the IMS German
Festival package and in file COPYING.bomp for the BOMP lexicon.
In order to run IMS German Festival you will need the Festival 2.5
(speech_tools and festival itself) already installed and running on
your machine. Get the sources from
http://www.festvox.org/festival/ and follow the
installation instructions. Test your installation by synthesizing an
English text. In case of problems please refer to the home page and
mailing lists available in Edinburgh.
Once Festival is running get the following four packages from the IMS
The sources and scheme file for the IMS German Festival
The BOMP lexicon provided by the IKP, University of Bonn,
with some modifications by us.
Unpack all the packages in the Festival base directory (this is where
you already have the two subdirectories speech_tools/ and festival/)
Additionally you will need the MBROLA engine and at least one voice
from the MBROLA project.
Install the the MBROLA engine in directory of your choice
(e.g. /usr/local/MBROLA/) and the voices in subdirectories below.
We will assume that you have an Unix machine although everything
should also work on Windows.
Add the following line to the Festival config file festival/config/config:
ALSO_INCLUDE += ims_german_text
Add the MBROLA directory with the mbrola binary (here we suppose it is
/usr/local/MBROLA/) and the path to the actual databases to the file
festival/lib/sitevars.scm (create the file if it does not exist):
(set! mbrola-path "/usr/local/MBROLA/Bin/")
(set! mbrola_database_path "/usr/local/MBROLA-voices/data/")
(set! mbrola_progname (string-append mbrola-path "mbrola -e"))
Add a line to the file festival/lib/siteinit.scm that loads the German
modules into Festival:
The IMS German Festival contains all modules necessary for
Text-to-Speech conversion. More explicitly this involves linguistic
text analysis (modules tokenization, token-pos, token-to-word)
grapheme-to-phoneme conversion (lexicon-lookup, letter-to-sound rules),
prosody (phrasing, intonation, duration, intonation targets) and
waveform synthesis (carried out by the MBROLA engine through an
For more background information about the IMS German Festival consult
the manual festival/doc/german/germanfestival-manual.ps or
To synthesize German text start festival
choose a German voice, e.g.
synthesize a text:
festival> (SayText "Hallo Welt.")
and enjoy listening to it.
You can also synthesize the German introduction text by typing
If you want to use a German voice with the text2wave script,
provide the scheme command to load the voice via -eval, e.g.
shell> echo hallo | $FESTDIR/examples/text2wave -o hallo.wav -eval '(voice_german_de2_os)'
Keep updated by regularly checking the IMS German Festival homepage
If you have questions or problems regarding Festival itself, please
post them to the festival mailing list
(more info on the Festival home page
If you encounter problems using German within Festival please mail to
the following address: