Exemple #1
0
    def default_test_prop():
        """Return the default proportion for the test set

        Try to read from configuration file, else return 0.5"""
        try:
            return float(config.get('split', 'default-test-proportion'))
        except ConfigParser.NoOptionError:
            return 0.5
Exemple #2
0
def kaldi_path():
    """Return an environment for running the abkhazia recipes"""
    env = os.environ

    env['LC_ALL'] = 'C'  # for expected sorting and joining behaviour

    kaldiroot = config.get('kaldi', 'kaldi-directory')
    kaldisrc = os.path.join(kaldiroot, 'src')

    targets = ('bin', 'featbin', 'fgmmbin', 'fstbin', 'gmmbin', 'latbin',
               'nnetbin', 'nnet2bin', 'sgmmbin', 'lmbin', 'kwsbin',
               'ivectorbin', 'online2bin', 'sgmm2bin')
    kaldibin = ':'.join((os.path.join(kaldisrc, target) for target in targets))

    fstbin = os.path.join(kaldiroot, 'tools', 'openfst', 'bin')

    platform = 'macosx' if os.uname()[0] is 'Darwin' else 'i686-m64'
    lmbin = ':'.join([
        os.path.join(kaldiroot, 'tools', 'irstlm', 'bin'),
        os.path.join(kaldiroot, 'tools', 'srilm', 'bin'),
        os.path.join(kaldiroot, 'tools', 'srilm', 'bin', platform),
        os.path.join(kaldiroot, 'tools', 'sctk', 'bin')
    ])

    if 'PATH' not in env:
        # PATH isn't in the environment, should not occur
        env['PATH'] = ':'.join([kaldibin, fstbin, lmbin])
    else:
        path = ':'.join([kaldibin, fstbin, lmbin])
        if path not in env['PATH']:
            env['PATH'] += ':' + path

    env['IRSTLM'] = os.path.join(kaldiroot, 'tools', 'irstlm')

    # was a bug -> error while loading shared libraries: libfstscript.so.1
    fstlib = os.path.join(kaldiroot, 'tools', 'openfst', 'lib')
    libfstscript = os.path.join(fstlib, 'libfstscript.so.1')
    assert os.path.exists(libfstscript)

    if 'LD_LIBRARY_PATH' not in env:
        env['LD_LIBRARY_PATH'] = fstlib
    else:
        if fstlib not in env['LD_LIBRARY_PATH']:
            env['LD_LIBRARY_PATH'] += ':' + fstlib

    return env
    def __init__(self, corpus, recipe_dir,
                 name='recipe', log=logger.null_logger()):
        # filter out short utterances
        self.corpus = corpus.subcorpus(
            self._desired_utterances(corpus), validate=False)

        # init the recipe directory, create it if needed
        self.recipe_dir = recipe_dir
        if not os.path.isdir(self.recipe_dir):
            os.makedirs(self.recipe_dir)

        self.name = name
        self.log = log

        # init the path to kaldi
        self.kaldi_root = config.get('kaldi', 'kaldi-directory')

        # init the path to abkhazia/share
        self.share_dir = pkg_resources.resource_filename(
            pkg_resources.Requirement.parse('abkhazia'), 'abkhazia/share')
Exemple #4
0
 def _write_cmd_script(script):
     with open(script, 'w') as stream:
         for cmd in ('train', 'decode', 'highmem'):
             key = '{}-cmd'.format(cmd)
             value = config.get('kaldi', key)
             stream.write('export {}={}\n'.format(key, value))