Python implementation of audformat
Project description
Specification and reference implementation of audformat.
audformat stores media data, such as audio or video, together with corresponding annotations in a pre-defined way. This makes it easy to combine or replace databases in machine learning projects.
An audformat database is a folder that contains media files together with a header YAML file and one or several files storing the annotations. The database is represented as an audformat.Database object and can be loaded with audformat.Database.load() or written to disk with audformat.Database.save().
Have a look at the installation and usage instructions and the format specifications as a starting point.
Changelog
All notable changes to this project will be documented in this file.
The format is based on Keep a Changelog, and this project adheres to Semantic Versioning.
Version 0.12.1 (2021-11-17)
Added: audformat.utils.hash()
Added: audformat.utils.expand_file_path()
Added: audformat.utils.replace_file_extension()
Changed: use yaml.CLoader for faster header reading
Version 0.12.0 (2021-11-10)
Added: as_segmented, allow_nat, root, num_workers arguments to audformat.Table.get()
Added: as_segmented, allow_nat, root, num_workers arguments to audformat.Column.get()
Added: files_duration argument to audformat.utils.to_segmented_index()
Added: audformat.Database.files_duration()
Changed: changed default value of load_data argument in audformat.Database.load() to False
Changed: speed up audformat.Database.files and audformat.Database.segments
Fixed: re-add support for pandas>=1.3
Version 0.11.6 (2021-08-20)
Added: support for Python 3.9
Fixed: speed up audformat.utils.union()
Fixed: audformat.Column.set() with pd.Series and np.array for a scheme with fixed labels and containing NaN values
Version 0.11.5 (2021-08-09)
Removed: duration scheme and column from conventions and emodb example
Version 0.11.4 (2021-08-05)
Added: custom BadKeyError when key is not found
Changed: limit to pandas <1.3 until it works again for newer pandas versions
Changed: remove the <1.0.0 limit for audiofile as a stable release is available and the API has not changed
Version 0.11.3 (2021-06-10)
Added: audformat.utils.duration
Fixed: description of audformat.Database.is_portable in documentation
Version 0.11.2 (2021-05-12)
Added: audformat.utils.join_schemes
Version 0.11.1 (2021-05-11)
Added: Database.is_portable
Added: copy_media argument to Database.update()
Changed: remove root argument from testing.create_audio_files() and instead use Database.root
Fixed: utils.concat() converts to nullable dtype
Fixed: utils.concat() returns DataFrame if input contains at least one DataFrame
Version 0.11.0 (2021-05-06)
Note: tables stored from this version upwards cannot be loaded with older versions
Added: Database.root
Added: utils.join_labels()
Added: Scheme.replace_labels()
Changed: set dependency to pandas>=1.1.5
Changed: do not compress pickled table files
Version 0.10.2 (2021-04-22)
Changed: allow_nat argument to utils.to_segmented_index()
Version 0.10.1 (2021-03-31)
Fixed: audformat.assert_index() checks for correct dtypes
Version 0.10.0 (2021-03-18)
Added: audformat.Database.update()
Added: audformat.Table.update()
Added: overwrite argument to audformat.utils.concat()
Changed: result of audformat.Table.__add__() is no longer assigned to a audformat.Database
Version 0.9.8 (2021-02-23)
Added: audformat.Database.license
Added: audformat.Database.license_url
Added: audformat.Database.author
Added: audformat.Database.organization
Added: audformat.utils.intersect() for index objects
Added: audformat.utils.union() for index objects
Changed: Database.load() raises error if table file missing
Changed: forbid duplicates in audformat conform indices
Fixed: audformat.Table.__add__() returned wrong values for some index combinations
Version 0.9.7 (2021-02-01)
Added: update_other_formats argument to audformat.Table.save() to make sure existing files in other formats are updated as well
Changed: use round_trip argument when loading CSV files to ensure dataframes are equal after storing and loading again
Version 0.9.6 (2021-01-28)
Fixed: implemented audformat.Database.__eq__ and return True for identical databases
Version 0.9.5 (2021-01-14)
Changed: use nullable Pandas’ type 'boolean' for bool schemes
Fixed: Scheme.draw() generates boolean values if scheme is bool
Version 0.9.4 (2021-01-11)
Changed: add arguments num_workers and verbose to audformat.Database.load()
Version 0.9.3 (2021-01-07)
Fixed: avoid sphinx syntax in CHANGELOG
Version 0.9.2 (2021-01-07)
Changed: add arguments num_workers and verbose to audformat.Database.drop_files(), audformat.Database.map_files(), audformat.Database.pick_files(), audformat.Database.save()
Changed: audformat.segmented_index() support int and float, which will be interpreted as seconds
Fixed: audformat.utils.to_segmented_index() returns correct index type for NaT
Version 0.9.1 (2020-12-21)
Fixed: add column name to HTML Series output in docs
Fixed: removed mentioning of NotConformToUnifiedFormat error and RedundantArgumentError error
Fixed: add missing errors to docstring of audformat.Table.set() and audformat.Column.set()
Version 0.9.0 (2020-12-18)
Added: initial release public release
Release history Release notifications | RSS feed
Download files
Download the file for your platform. If you're not sure which to choose, learn more about installing packages.
Source Distribution
Built Distribution
File details
Details for the file audformat-0.12.1.tar.gz.
File metadata
- Download URL: audformat-0.12.1.tar.gz
- Upload date:
- Size: 77.2 kB
- Tags: Source
- Uploaded using Trusted Publishing? No
- Uploaded via: twine/3.6.0 importlib_metadata/4.8.2 pkginfo/1.7.1 requests/2.26.0 requests-toolbelt/0.9.1 tqdm/4.62.3 CPython/3.10.0
File hashes
| Algorithm | Hash digest | |
|---|---|---|
| SHA256 | d29898fcf0b2157babcbc424ea7c6c73575d0f544342ed109f8cc75cef75e3c7 |
|
| MD5 | ed9ca49c834dc48a89236cf6de708329 |
|
| BLAKE2b-256 | a15391a000cbcacda6f7705c821c548d65e54148801c957500b3ecd01d045d0d |
File details
Details for the file audformat-0.12.1-py3-none-any.whl.
File metadata
- Download URL: audformat-0.12.1-py3-none-any.whl
- Upload date:
- Size: 46.8 kB
- Tags: Python 3
- Uploaded using Trusted Publishing? No
- Uploaded via: twine/3.6.0 importlib_metadata/4.8.2 pkginfo/1.7.1 requests/2.26.0 requests-toolbelt/0.9.1 tqdm/4.62.3 CPython/3.10.0
File hashes
| Algorithm | Hash digest | |
|---|---|---|
| SHA256 | c97f034622d45328ffb8b662a38e70d116023b396c64c064143bc722a64b4718 |
|
| MD5 | 4c1c22a2dec8bb6ac2f854b345a975d7 |
|
| BLAKE2b-256 | 72d0e14898d1d3ca929b2eea66ef3064479f375914be0b1ba4e484714fa7fae8 |
