ELAN2split,
(C) 2015-2016 by Damir Cavar:
This is a free and open curse implementation of a segmenter for ELAN files written in C++11 and
platform independent. Using this tool one can select any tier in an
ELAN Annotation File and generate file-pairs of
audio segment and corresponding transcription all annotated time intervals. These segments of audio and
transcription can be used to train common Forced Alignment tools, e.g. the
Penn Forced Aligner,
the Prosodylab-Aligner, or
or the MAUS
Segmenter.
A set of binaries is available on the
Bitbucket project page.
The code of ELAN2split is free and open source. SoX
is a prerequisite. Please contact us, if you need a binary
for your project or system.
(back to Contents)
Espeak language models for Text-to-Speech, (C) 2015 by
Lwin Moe,
Andrew Lamont,
Damir Cavar,
Malgorzata E. Cavar:
This is a collection of Espeak language models for
Text-to-Speech. These models were developed to serve
as plugins in the Praat-based Forced Alignment
implementation. The Praat-based implementation uses
Espeak
to generate audio from text and forced align transcriptions. We develop models for different
low-resourced
and endangered languages to serve in Praat-based forced
alignment.
TreebankParser SA is a small
tool to extract rules from treebanks that use the Penn Treebank notation.
It generates relative or absolute frequency profiles for the extracted rules. It is part of the
Free Linguistic Environment
(FLE) project.