wenwanchen/kaldi-io-for-python

Python functions for reading kaldi data formats. Useful for rapid prototyping with python.

kaldi-io-for-python

'Glue' code connecting kaldi data and python.

Supported data types

• vector (integer)
• Vector (float, double)
• Matrix (float, double)
• Posterior (posteriors, nnet1 training targets, confusion networks, ...)

Examples

Reading feature scp example:

import kaldi_io
for key,mat in kaldi_io.read_mat_scp(file):
  ...

Writing feature ark to file/stream:

import kaldi_io
with open(ark_file,'wb') as f:
  for key,mat in dict.iteritems():
    kaldi_io.write_mat(f, mat, key=key)

Writing features as 'ark,scp' by pipeline with 'copy-feats':

import kaldi_io
ark_scp_output='ark:| copy-feats --compress=true ark:- ark,scp:data/feats2.ark,data/feats2.scp'
with kaldi_io.open_or_fd(ark_scp_output,'wb') as f:
  for key,mat in dict.iteritems():
    kaldi_io.write_mat(f, mat, key=key)

Install

• from pypi: python -m pip --user install kaldi_io
• from sources:
  • git clone https://github.com/vesis84/kaldi-io-for-python.git <kaldi-io-dir>
  • python setup.py install (default python)
• for local development use: export PYTHONPATH=${PYTHONPATH}:<kaldi-io-dir> in $HOME/.bashrc

Note: it is recommended to set $KALDI_ROOT in your $HOME/.bashrc as export KALDI_ROOT=<some_kaldi_dir>, so you can read/write using pipes which contain kaldi binaries.

License

Apache License, Version 2.0 ('LICENSE-2.0.txt')

Contact

• If you have an extension to share, please create a pull request.
• For feedback and suggestions, please create a GitHub 'Issue' in the project.
• For the positive reactions =) I am also reachable by email: [email protected]