timit-preprocessor extract mfcc vectors and phones from TIMIT dataset for advanced use on speech recognition.
The TIMIT corpus of read speech is designed to provide speech data for acoustic-phonetic studies and for the development and evaluation of automatic speech recognition systems.
More information on website or Wiki
The instructions and scripts used here are built upon timit-preprocessor.
make_dataset.py
relies on kaldi-io-for-python
You must have downloaded the TIMIT dataset. You must have a compiled version of Kaldi.
Note that to install Kaldi first by following the instructions in INSTALL
.
(1)
go to tools/ and follow INSTALL instructions there.(2) go to src/ and follow INSTALL instructions there.
After running the scripts instructed by INSTALL
in tools/
, there will be reminder as followed. Go and run it.
Kaldi Warning: IRSTLM is not installed by default anymore. If you need IRSTLM, use the script
extras/install_irstlm.sh
- source the python interpreter matching the requirement.txt file.
$ source ../../pyenv/bin/activate
- Edit the default of variables
KALDI_ROOT
,TIMIT_ROOT
,DATA_OUT
in the Makefile to match your installation. You can also leave the default as is and usemake
with location arguments.
$ make KALDI_ROOT=abc/kaldi TIMIT_ROOT=abc/timit DATA_OUT=abc/out ...
- Run the following commands (here without location arguments):
$ make convert
$ make -j 4
Note 1: noisy .wav
files will be created alongside timit clean ones.
Note 2: In case of errors, display the remaining steps:
$ make -n
and try to debug them one by one.
Note 3: For serious problems you can always contact us in the [issues] section.
Some codes of the TIMIT Preprocessor are from the following repo: TIMIT Preprocessor