SailAlign Long speech-text alignment can facilitate large-scale study of rich spoken language resources that have recently become widely accessible, e.g., collections of audio books, or multimedia documents. SailAlign is an open-source software toolkit for robust long speech-text alignment implementing an adaptive, iterative speech recognition and text alignment scheme that allows for the processing of very long (and possibly noisy) audio and is robust to transcription errors. The software package was originally presented at VLSP-2011: A. Katsamanis, M. Black, P. Georgiou, L. Goldstein and S. Narayanan, SailAlign: Robust long speech-text alignment, in Proc. of Workshop on New Tools and Methods for Very-Large Scale Phonetics Research, Jan. 2011. INSTALLATION To install this module, run the following command while inside the SailAlign distribution folder. The module can currently only work on linux - like operating systems. It has been tested in the following architectures: linux_i686, linux_x86_64 Compiled HTK binaries, including HDecode have to be available in your system, e.g., in the folder $HTK_BIN. You may need to be connected to the internet to install dependencies : 1) cpan Module::Build This will install the Module::Build perl package from CPAN. You may skip this step if you have the package already installed. It is used for the installation of the SailAlign package. If you haven't run the command cpan before you may have to go through a setup process. You can just select the default options when user input is required. 2) perl Build.PL This command configures the building process. Among others it will try to detect dependencies that are not installed in your system. Ignore any messages for missing dependencies in the beginning. Provide the folder $HTK_BIN when asked. Internet connection is required at this point to download the acoustic models. 3) sudo ./Build installdeps Run this command if you got a message for missing dependencies in the previous step. 3) ./Build This command will build the module. 4) sudo ./Build install This command will install the module. The following perl scripts will be added to your path: sail_align, sail_align_parallel TESTING To test that the module has been properly installed you may run the following from inside the distribution folder: >> sail_align -i support/data/timit_5.wav -t support/data/timit_5.txt -w support/test/local \ -e timit_sample_test -c config/timit_alignment.cfg This will run the alignment algorithm for the audio file timit_5.wav and the corresponding transcription timit_5.txt. The results will be in the working directory 'support/test/local/timit_sample_test'. You may check the tutorial in the 'docs' folder for additional details. For spanish: >> sail_align -i support/data/spanish/voxforge_spanish.wav -t support/data/spanish/voxforge_spanish.txt -w support/test/local_spanish \ -e spanish_test -c config/spanish_alignment.cfg For greek: >> sail_align -i support/data/greek/greeksyn_05.wav -t support/data/greek/greeksyn_05.txt -w support/test/local_greek \ -e greek_test -c config/greek_alignment.cfg FIXES In case you get a warning like the following, there is no need to worry: "Ambiguous call resolved as CORE::read(), qualify as such or use & at (eval 18) line 4. Subroutine Audio::Wav::Read::read redefined at /usr/lib/perl5/site_perl/5.10.0/Audio/Wav/Read.pm line 316." It is due to a minor bug in the Audio::Wav external dependency and it does not affect sail_align's result. To avoid it you may apply the following patch: sudo patch /usr/lib/perl5/site_perl/5.10.0/Audio/Wav/Read.pm < support/patches/audio_wav_read.patch SUPPORT AND DOCUMENTATION You may find additional documentation inside the docs subfolder, including the paper on SailAlign presented at VLSP-2011 and a relevant tutorial that was given there as well. Contact Nassos Katsamanis for technical support. LICENSE AND COPYRIGHT Copyright (C) 2010-2013 Athanasios Katsamanis (http://sipi.usc.edu/~nkatsam) This program is free software; you can redistribute it and/or modify it under the terms of the GNU General Public License as published by the Free Software Foundation; version 2 dated June, 1991 or at your option any later version. SRILM binaries are only included in the distribution for convenience and you may find the corresponding license, srilm-license, in the folder LICENSES. This program is distributed in the hope that it will be useful, but WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License for more details. A copy of the GNU General Public License is available in the source tree; if not, write to the Free Software Foundation, Inc., 59 Temple Place - Suite 330, Boston, MA 02111-1307, USA.
nassosoassos/sail_align
SailAlign is an open-source software toolkit for robust long speech-text alignment implementing an adaptive, iterative speech recognition and text alignment scheme that allows for the processing of very long (and possibly noisy) audio and is robust to transcription errors. It is mainly written as a perl library but its functionality also depends on freely available software, namely HTK, srilm and sclite.
Perl