Merge pull request #32 from NickleDave/add-textgrid

Add textgrid

setup.py

@@ -29,6 +29,7 @@
     'crowsetta.format': [
         'notmat = crowsetta.notmat',
         'koumura = crowsetta.koumura',
+        'textgrid = crowsetta.textgrid',
     ]
 }
 

src/crowsetta/__init__.py

@@ -1,5 +1,3 @@
-from . import notmat
-from . import koumura
 from .transcriber import Transcriber
 from .segment import Segment
 from .sequence import Sequence
@@ -8,3 +6,7 @@
 from . import data
 from . import formats
 
+# built-in formats
+from . import notmat
+from . import koumura
+from . import textgrid

src/crowsetta/notmat.py

@@ -2,7 +2,6 @@
 produced by evsonganaly GUI
 """
 import os
-from pathlib import Path
 
 import numpy as np
 import scipy.io
@@ -11,29 +10,7 @@
 from .sequence import Sequence
 from .csv import seq2csv
 from .meta import Meta
-
-
-def _parse_file(file):
-    """helper function that parses/validates value for file argument;
-    puts a single string or Path into a list to iterate over it (cheap hack
-    that lets functions accept multiple types), and checks list to make sure
-    all types are consistent
-    """
-    if type(file) == str or type(file) == Path:
-        # put in a list to iterate over
-        file = [file]
-
-    for a_file in file:
-        if type(a_file) == str:
-            if not a_file.endswith('.not.mat'):
-                raise ValueError("all filenames in .not.mat must end with '.not.mat' "
-                                 f"but {a_file} does not")
-        elif type(a_file) == Path:
-            if not a_file.suffixes == ['.not', '.mat']:
-                raise ValueError("all filenames in .not.mat must end with '.not.mat' "
-                                 f"but {a_file} does not")
-
-    return file
+from .validation import _parse_file
 
 
 def notmat2seq(file,
@@ -76,7 +53,7 @@ def notmat2seq(file,
     due to floating point error, e.g. when loading .not.mat files and then sending them to
     a csv file, the result should be the same on Windows and Linux
     """
-    file = _parse_file(file)
+    file = _parse_file(file, extension='.not.mat')
 
     if abspath and basename:
         raise ValueError('abspath and basename arguments cannot both be set to True, '
@@ -158,7 +135,7 @@ def notmat2csv(file, csv_filename, abspath=False, basename=False):
     -------
     None
     """
-    file = _parse_file(file)
+    file = _parse_file(file, extension='.not.mat')
 
     if abspath and basename:
         raise ValueError('abspath and basename arguments cannot both be set to True, '

src/crowsetta/textgrid.py

@@ -0,0 +1,154 @@
+"""module for loading Praat TextGrid files into Sequences
+
+uses the Python library textgrid
+https://github.com/kylebgorman/textgrid
+a version is distributed with this code (../textgrid) under MIT license
+https://github.com/kylebgorman/textgrid/blob/master/LICENSE
+"""
+import os
+
+import numpy as np
+from textgrid import TextGrid, IntervalTier
+
+from .sequence import Sequence
+from .meta import Meta
+from .csv import seq2csv
+from .validation import _parse_file
+
+
+def textgrid2seq(file,
+                 abspath=False,
+                 basename=False,
+                 round_times=True,
+                 decimals=3,
+                 intervaltier_ind=0,
+                 audio_ext='wav',
+                 ):
+    """convert Praat Textgrid file into a Sequence
+
+    Parameters
+    ----------
+    file : str, Path
+        filename of a .TextGrid annotation file, created by Praat.
+    abspath : bool
+        if True, converts filename for each audio file into absolute path.
+        Default is False.
+    basename : bool
+        if True, discard any information about path and just use file name.
+        Default is False.
+    round_times : bool
+        if True, round onsets_s and offsets_s.
+        Default is True.
+    decimals : int
+        number of decimals places to round floating point numbers to.
+        Only meaningful if round_times is True.
+        Default is 3, so that times are rounded to milliseconds.
+    intervaltier_ind : int
+        index of IntervalTier in TextGrid file from which annotations
+        should be taken. Default it 0, i.e. the first IntervalTier.
+        Necessary in cases where files have multiple IntervalTiers.
+        Currently there is only support for converting a single IntervalTier
+        to a single Sequence.
+    audio_ext : str
+        extension of audio file associated with TextGrid file. Used to
+        determine the filename of the audio file associated with the
+        TextGrid file. Default is 'wav'.
+
+    Returns
+    -------
+    seq : crowsetta.Sequence or list of Sequence
+        each Interval in the first IntervalTier in a TextGrid file
+        will become one segment in a sequence.
+    """
+    file = _parse_file(file, extension='.TextGrid')
+    seq = []
+    for a_textgrid in file:
+        tg = TextGrid.fromFile(a_textgrid)
+
+        intv_tier = tg[intervaltier_ind]
+        if type(intv_tier) != IntervalTier:
+            raise ValueError(f'Index specified for IntervalTier was {intervaltier_ind}, '
+                             f'but type at that index was {type(intv_tier)}, not an IntervalTier')
+
+        audio_filename = a_textgrid.replace('TextGrid', audio_ext)
+        if abspath:
+            audio_filename = os.path.abspath(audio_filename)
+        elif basename:
+            audio_filename = os.path.basename(audio_filename)
+
+        intv_tier = tg[0]
+        onsets_s = []
+        offsets_s = []
+        labels = []
+
+        for interval in intv_tier:
+            labels.append(interval.mark)
+            onsets_s.append(interval.minTime)
+            offsets_s.append(interval.maxTime)
+
+        labels = np.asarray(labels)
+        onsets_s = np.asarray(onsets_s)
+        offsets_s = np.asarray(offsets_s)
+
+        # do this *after* converting onsets_s and offsets_s to onsets_Hz and offsets_Hz
+        # probably doesn't matter but why introduce more noise?
+        if round_times:
+            onsets_s = np.around(onsets_s, decimals=decimals)
+            offsets_s = np.around(offsets_s, decimals=decimals)
+
+        textgrid_seq = Sequence.from_keyword(file=audio_filename,
+                                             labels=labels,
+                                             onsets_s=onsets_s,
+                                             offsets_s=offsets_s)
+        seq.append(textgrid_seq)
+
+    if len(seq) == 1:
+        return seq[0]
+    else:
+        return seq
+
+
+def textgrid2csv(file, csv_filename, abspath=False, basename=False):
+    """saves annotation from TextGrid file(s) in a comma-separated values
+    (csv) file, where each row represents one syllable from one
+    .not.mat file.
+
+    Parameters
+    ----------
+    file : str, Path, or list
+        if list, list of strings or Path objects pointing to TextGrid files
+    csv_filename : str
+        name for csv file that is created
+
+    The following two parameters specify how file names for audio files are saved. These
+    options are useful for working with multiple copies of files and for reproducibility.
+    Default for both is False, in which case the filename is saved just as it is passed to
+    this function.
+    abspath : bool
+        if True, converts filename for each audio file into absolute path.
+        Default is False.
+    basename : bool
+        if True, discard any information about path and just use file name.
+        Default is False.
+
+    Returns
+    -------
+    None
+    """
+    file = _parse_file(file, extension='.TextGrid')
+
+    if abspath and basename:
+        raise ValueError('abspath and basename arguments cannot both be set to True, '
+                         'unclear whether absolute path should be saved or if no path '
+                         'information (just base filename) should be saved.')
+
+    seq = textgrid2seq(file)
+    seq2csv(seq, csv_filename, abspath=abspath, basename=basename)
+
+
+meta = Meta(
+    name='textgrid',
+    ext='TextGrid',
+    to_seq=textgrid2seq,
+    to_csv=textgrid2csv,
+)

src/crowsetta/validation.py

@@ -1,17 +1,10 @@
 """utilities for input validation.
 
-Adapted from scikit-learn under BSD 3 License
+Some utilities adapted from scikit-learn under BSD 3 License
 https://github.com/scikit-learn/scikit-learn/blob/master/sklearn/utils/validation.py
-
-Original Authors: Olivier Grisel
-                  Gael Varoquaux
-                  Andreas Mueller
-                  Lars Buitinck
-                  Alexandre Gramfort
-                  Nicolas Tresegnie
 """
-import warnings
 import numbers
+from pathlib import Path
 
 import numpy as np
 
@@ -69,3 +62,25 @@ def column_or_row_or_1d(y):
         return np.ravel(y)
     else:
         raise ValueError("bad input shape {0}".format(shape))
+
+
+def _parse_file(file, extension):
+    """helper function that parses/validates value for file argument;
+    puts a single string or Path into a list to iterate over it (cheap hack
+    that lets functions accept multiple types), and checks list to make sure
+    all types are consistent
+    """
+    if type(file) == str or type(file) == Path:
+        # put in a list to iterate over
+        file = [file]
+
+    for a_file in file:
+        # cast to string (if it's not already, e.g. it's a Path)
+        # so we can use .endswith() to compare extensions
+        # (because using Path.suffixes() would require too much special casing)
+        a_file = str(a_file)
+        if not a_file.endswith(extension):
+            raise ValueError(f"all filenames in 'file' must end with '{extension}' "
+                             f"but {a_file} does not")
+
+    return file

src/textgrid/AUTHORS

@@ -0,0 +1,5 @@
+Max Bane <bane@uchicago.edu>, University of Chicago
+Kyle Gorman <kgorman@ling.upenn.edu>, University of Pennsylvania
+Morgan Sonderegger <morgan@cs.uchicago.edu>
+
+KG designed this for personal use in 2007, circulated it in 2008, and turned the original methods into Python classes in 2009. MS and MB began fixing bugs and contributing functions for particular use cases in 2011. 

src/textgrid/LICENSE

@@ -0,0 +1,7 @@
+Copyright (c) 2011-2014 Kyle Gorman, Max Bane, Morgan Sonderegger
+
+Permission is hereby granted, free of charge, to any person obtaining a copy of this software and associated documentation files (the "Software"), to deal in the Software without restriction, including without limitation the rights to use, copy, modify, merge, publish, distribute, sublicense, and/or sell copies of the Software, and to permit persons to whom the Software is furnished to do so, subject to the following conditions:
+
+The above copyright notice and this permission notice shall be included in all copies or substantial portions of the Software.
+
+THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.

src/textgrid/README.md

@@ -0,0 +1,27 @@
+textgrid.py
+===========
+
+Python classes for Praat TextGrid and TextTier files (and HTK .mlf files)
+
+Kyle Gorman <kylebgorman@gmail.com> and contributors (see commit history).
+
+How to cite:
+------------
+
+While you don't have to, if you want to cite textgrid.py in a publication, include a footnote link to the source:
+
+    http://github.com/kylebgorman/textgrid.py/
+
+How to install:
+---------------
+
+The code can be placed in your working directory or in your `$PYTHONPATH`,  and then imported in your Python script. You also can install it via `pip`, like so:
+
+    pip install textgrid
+
+(if you're not working in a virtualenv, you may need to do this with `sudo`.)
+
+Synopsis:
+---------
+
+See the docstrings in `textgrid.py`

src/textgrid/__init__.py

@@ -0,0 +1 @@
+from .textgrid import TextGrid, MLF, IntervalTier, PointTier, Interval, Point

src/textgrid/exceptions.py

@@ -0,0 +1,3 @@
+
+class TextGridError(Exception):
+    pass