Skip to content

Latest commit

 

History

History
65 lines (44 loc) · 3 KB

CHANGELOG.md

File metadata and controls

65 lines (44 loc) · 3 KB
Raw

Changelog

version 0.3.0 (19 Nov 2021):

  1. Changed module naming. nnAudio.Spectrogram will be replaced by nnAudio.features in the future releases. Currently, various spectrogram types are accessible via both methods.

version 0.2.6 (02 Sep 2021):

  1. Add relu() to the nonlinear_func in CFP() to prevent negative values #105.

version 0.2.5 (06 Aug 2021):

  1. Incorrect inverse STFT calculation is fixed #100.
  2. Add more test cases in unit test
  3. Refactor STFT and iSTFT

version 0.2.5 (06 Aug 2021):

  1. Incorrect inverse STFT calculation is fixed #100.
  2. Add more test cases in unit test
  3. Refactor STFT and iSTFT This version can be obtained via pip install nnAudio==0.2.5.

version 0.2.4 (11 June 2021):

  1. CQT2010 bug has been fixed #85.
  2. Provide a wider support for scipy versions using from scipy.fftpack import fft in utils.py
  3. Documentation error for STFT has been fixed #90

This version can be obtained via pip install nnAudio==0.2.4.

version 0.2.3 (10 June 2021):

Broken package, please ignore this one.

version 0.2.2 (1 March 2021):

Added filter scale support to various version of CQT classes as requested in #54. Different normalization methods are also added to the forward() method as normalization_type under each CQT class. A bug is discovered in CQT2010, the output is problematic #85.

To use this version, do pip install nnAudio==0.2.2.

version 0.2.1 (15 Jan 2021):

Fixed bugs #80, #82, and fulfilled request #83. nnAudio version can be checked with nnAudio.__version__ inside python now. Added two more spectrogram types Gammatonegram() and Combined_Frequency_Periodicity().

To use this version, do pip install nnAudio==0.2.1.

version 0.2.0 (8 Nov 2020):

Now it is possible to do stft_layer.to(device) to move the spectrogram layers between different devices. No more device argument when creating the spectrogram layers.

To use this version, do pip install nnAudio==0.2.0.

version 0.1.5:

Much better iSTFT and Griffin-Lim. Now Griffin-Lim is a separated PyTorch class and requires torch >= 1.6.0 to run. STFT has also been refactored and it is less memory consuming now.

To use this version, do pip install nnAudio==0.1.5.

version 0.1.4a0:

Finalized iSTFT and Griffin-Lim. They are now more accurate and stable.

version 0.1.2.dev3:

Add win_length to STFT so that it has the same funcationality as librosa.

version 0.1.2.dev2:

Fix bugs where the inverse cannot be done using GPU. And add a separated iSTFT layer class

version 0.1.2.dev1:

Add Inverse STFT and Griffin-Lim. They are still under development, please use with care.

version 0.1.1 (1 June 2020):

Add MFCC