Skip to content

Latest commit

 

History

History
211 lines (145 loc) · 6.21 KB

CHANGELOG.md

File metadata and controls

211 lines (145 loc) · 6.21 KB
Raw

Changelog

0.5.0 - 2021-12-09

Official Open JOSS reviewed version.

Bugs

  • Fix bug of name conflict while loading chord model.

0.4.2 - 2021-11-16

Accumulated release. Various improvements and bug fix. See details below.

Feature

  • Migrate checkpoints from private Google Drive to Github release. See here
  • Replace opencv

Dependency

  • Upgrade Tensorflow version to 2.5.0 for Nvidia 30 series GPU compatibility.
  • Upgrade Spleeter version to 2.3.0 for new TF version compatibility.
  • Replace Opencv with PIL for drum feature resizing and remove opencv from the dependency.

Enhancement

  • Simplify model loading mechanism by unifying the all checkpoint format to use TF format.
  • Lazy import extraction functions to boost loading time.
  • Change the order of Dockerfile commands for better utilizing cache.

Documentation

  • Add notice about compatibility issue of running on certain CPU architecture.
  • Add explaination about enabling auto completion.
  • Rephrase sentences in paper according to JOSS review feedback.
  • Add explaination about installing development dependencies.
  • Use pepy as the alternative source for 'download' state badge.

Bugs

  • Fix bug of unable to find vocal contour checkpoint.
  • Fix bug of fail to custom layers of chord module.
  • Fix various unit tests bugs.
  • Fix minor linter errors.

0.4.1 - 2021-06-04

Hotfix version according to issue #23

Feature

  • Add a new piano transcription model and set it as the default model while using music module.

Bugs

  • Fix bug while parsing weight files in the checkpoint folder.

0.4.0 - 2021-05-31

Various improvements on music module and some critical bug fixes.

Enhancement

  • Improve the peak finding and thresholding strategy for more stable and better performance.
  • Modify the feeding strategy of feature slices with adjustable overlapping rate while making predictions.
  • Apply learning rate scheduler for music module.
  • Replace the usage of custom training loop of music module with the built-in TF .fit() function.

Bugs

  • Fix a critical bug of inference of music module that would lead to missing onsets.
  • Fix generation of pertubation of vocal module while training.

Documentation

  • Merge the demo page into master from build_doc branch.

0.3.4 - 2021-05-10

Hotifx version according to issue #19.

Bugs

  • Fix bug of treating numpy array as list while appending elements.

0.3.3 - 2021-05-07

Hotfix version according to issue #19.

Bugs

  • Fix column inconsistency of aggregate_f0_info and write_agg_f0_results.
  • Update version of dependencies according to the security alert.

0.3.2 - 2021-02-13

Enhancement

  • Move load_label functions of different datasets into dataset structure classes.
  • Add custom exception on fail downloading GD file due to access limit.
  • Add unit tests on parsing label files into shared intermediate format.

Bugs

  • Fix wrong access name of the dict in vocal midi inference function.
  • Fix bug of generating beat module training labels.

0.3.1 - 2021-01-18

Hotfix release of spleeter error.

Bugs

  • Call Spleeter in CLI mode instead of using python class.

0.3.0 - 2021-01-17

Release the beat module for symbolic domain beat transcription.

Features

  • Release beat module.
  • Add an example patch-cnn module for demonstrating the implementation progress.

Enhancement

  • Refactor the flow of chord module for parsing the feature and label files.
  • Modularize F0 information aggragation functions to utils.py and io.py.
  • Improve verbosity on fail to open hdf files.

Documentation

  • Re-arrange the side bar with an additional group of CLI.
  • Add custom CSS style for adjusting the width of audio and video elements.

Bugs

  • Fix Spleeter import errors after upgrading to v2.1.2.

0.2.0 - 2020-12-13

Vocal melody transcription in both frame- and note-level are live!

We release the modules for vocal melody transcription after a decent amount of effort. Now you can transcribe your favorite singing voice.

Features

  • Release vocal and vocal-contour submodules.

Enhancement

  • Improve chord transcription results by filtering out chord predictions with short duration.
  • Resolve the path for transcription output in a consistent way.

Documentation

  • Re-organize Quick Start and Tutorial pages to improve accessibility.
  • Move the section for development from README.md to CONTRIBUTING.md.

Bug Fix

  • Fix bugs of passing the wrong parameter to vamp for chroma feature extraction.

0.1.1 - 2020-12-01

Features

  • Add more supported datasets for download and process.
  • Supports to save checkpoints in .pb format with customized model checkpoint callback.

Enhancement

  • Huge refactor of constants.dataset. Improves reusability and add more useful common utilities.
  • Modularize common parts of app classes.
  • Construct base class of loading dataset samples. Reduce duplicate code and reuse the same functionalities.
  • Filter out messy Tensorflow warnings when using CLI.

Bug Fix

  • Resolved bugs of some function parameters not actually being used inside functions.
  • Fix CFP extraction down_fs don't actually work.

0.1.0 - 2020-11-16

Features

  • Add command for synthesizing MIDI file.
  • Provides colab for quick start now!

Enhancement

  • Lazy import application instance for avoiding pulling large amount of dependencies.
  • Group sub-commands into different sections when showing help message.

0.1.0-beta.2 - 2020-11-10

Enhancement

  • Better dealing with the input model path.
  • Better approach for resolving dataset path when given with "./".
  • Add documentation for Conda user for manually install omnizart.

Bug Fix

  • Fix wrong save path of checkpoints.
  • Fix installation script for not upgrading pip after activating virtual environment.

0.1.0-beta.1 - 2020-11-08

First release of omnizart CLI tool, as well as a python package.

Features

  • Multi-instrument transcription
  • Drum transcription
  • Chord transcription
  • Download datasets
  • Extract feature of datasets for each module
  • Train models for each module