crowsetta 4.0.0

A tool to work with any format for annotating vocalizations

---

a tool to work with any format for annotating vocalizations

crowsetta is a tool to work with any format for annotating vocalizations: speech, birdsong, mouse ultrasonic calls (insert your favorite animal vocalization here). The goal of crowsetta is to make sure that your ability to work with a dataset of vocalizations does not depend on your ability to work with any given format for annotating that dataset. What crowsetta gives you is not yet another format for annotation (I promise!); instead you get some nice data types that make it easy to work with any format: namely, Sequences made up of Segments.

    >>> from crowsetta import Segment, Sequence
    >>> a_segment = Segment.from_keyword(
    ...     label='a',
    ...     onset_ind=16000,
    ...     offset_ind=32000,
    ...     file='bird21.wav'
    ...     )
    >>> list_of_segments = [a_segment] * 3
    >>> seq = Sequence(segments=list_of_segments)
    >>> print(seq)
    Sequence(segments=[Segment(label='a', onset_s=None, offset_s=None, onset_ind=16000,
    offset_ind=32000, file='bird21.wav'), Segment(label='a', onset_s=None, offset_s=None,
    onset_ind=16000, offset_ind=32000, file='bird21.wav'), Segment(label='a', onset_s=None,
    offset_s=None, onset_ind=16000, offset_ind=32000, file='bird21.wav')])

You can load annotation from your format of choice into Sequences of Segments (most conveniently with the Transcriber, as explained below) and then use the Sequences however you need to in your program.

For example, if you want to loop through the Segments of each Sequences to pull syllables out of a spectrogram, you can do something like this, very Pythonically:

   >>> syllables_from_sequences = []
   >>> for a_seq in seq:
   ...     seq_dict = seq.to_dict()  # convert to dict with
   ...     spect = some_spectrogram_making_function(seq['file'])
   ...     syllables = []
   ...     for seg in seq.segments:
   ...         syllable = spect[:, seg.onset:seg.offset]  ## spectrogram is a 2d numpy array
   ...         syllables.append(syllable)
   ...     syllables_from_sequences.append(syllables)

As mentioned above, crowsetta provides you with a Transcriber that comes equipped with convenience functions to do the work of converting for you.

    from crowsetta import Transcriber
    scribe = Transcriber()
    seq = scribe.to_seq(file=notmat_files, format='notmat')

You can even easily adapt the Transcriber to use your own in-house format, like so:

    from crowsetta import Transcriber
    scribe = Transciber(user_config=your_config)
    scribe.to_csv(file_'your_annotation_file.mat',
                  csv_filename='your_annotation.csv')

Features

• convert annotation formats to Sequence objects that can be easily used in a Python program
• convert Sequence objects to comma-separated value text files that can be read on any system
• load comma-separated values files back into Python and convert to other formats
• easily use with your own annotation format

You might find it useful in any situation where you want to share audio files of song and some associated annotations, but you don't want to require the user to install a large application in order to work with the annotation files.

Getting Started

Installation

with pip

$ pip install crowsetta

with conda

$ conda install crowsetta -c conda-forge

Usage

To learn how to use crowsetta, please see the documentation at:
https://crowsetta.readthedocs.io/en/latest/index.html

Development Installation

Currently crowsetta is developed with conda. To set up a development environment:

$ conda create crowsetta-dev
$ conda create -n crowsetta-dev python=3.6 numpy scipy attrs
$ conda activate crowsetta-dev
$ $ pip install evfuncs koumura
$ git clone https://github.com/NickleDave/crowsetta.git
$ cd crowsetta
$ pip install -e .

Project Information

Background

crowsetta was developed for two libraries:

• hybrid-vocal-classifier https://github.com/NickleDave/hybrid-vocal-classifier
• vak https://github.com/NickleDave/vak

Testing relies on the Vocalization Annotation Formats Dataset which you may find useful if you need small samples of different audio files and associated annotation formats

• on Figshare: https://figshare.com/articles/Vocalization_Annotation_Formats_Dataset/8046920
• built from this GitHub repository: https://github.com/NickleDave/vocal-annotation-formats

Support

If you are having issues, please let us know.

• Issue Tracker: https://github.com/NickleDave/crowsetta/issues

Contribute

• Issue Tracker: https://github.com/NickleDave/crowsetta/issues
• Source Code: https://github.com/NickleDave/crowsetta

CHANGELOG

You can see project history and work in progress in the CHANGELOG

License

The project is licensed under the BSD license.

Citation

If you use crowsetta, please cite the DOI: