v1.6.5

Formal release of 1.6.5 which adds a choose_weighted distribution tuple for non-uniform sampling from a discrete set.

v1.6.5.rc0

• Added a new distirbution tuple: ("choose_weighted", list_of_options, probabilities), which supports weighted sampling: list_of_options[i] is chosen with probability probabilities[i].

v1.6.4

• Scaper.generate now accepts a new argument for controlling trade-off between speed and quality in pitch shifting and time stretching:
  • quick_pitch_time: if True, both time stretching and pitch shifting will be applied in quick mode, which is much faster but has lower quality.

v1.6.3

• Scaper.generate now accepts two new optional arguments for controlling audio clipping and normalization:
  • fix_clipping: if True and the soundscape audio is clipping, it will be peak normalized and all isolated events will be scaled accordingly.
  • peak_normalization: if True, sounscape audio will be peak normalized regardless of whether it's clipping or not and all isolated events will be scaled accordingly.
• All generate arguments are now documented in the scaper sandbox inside the JAMS annotation.
• Furthermore, we also document in the JAMS: the scale factor used for peak normalization, the change in ref_db, and the actual ref_db of the generated audio.

v1.6.2

• Switching from FFMpeg LUFS calculation to pyloudnorm for better performance: runtime is reduced by approximately 30%
• The loudness calculation between FFMpeg LUFS and pyloudnorm is slightly different so this version will generate marginally different audio data compared to previous versions: the change is not perceptible, but np.allclose() tests on audio from previous versions of Scaper may fail.
• This change updates the regression data for Scaper's regression tests.
• This release used soxbindings 1.2.2 and pyloudnorm 0.1.0.

v1.6.1

• Trimming now happens on read, rather than after read. This prevents the entire file from being loaded into memory. This is helpful for long source audio files.
• Since the audio processing pipeline has changed, this version will generate marginally different audio data compared to previous versions: the change is not perceptible, but np.allclose() tests on audio from previous versions of Scaper may fail.
• This change updates the regression data for Scaper's regression tests

v1.5.1

• Fixes a bug with fade in and out lengths are set to 0.
• This is the last version to support Python 2.7 and 3.4.

v1.6.0

• Scaper now uses soxbindings instead of sox when installing on Linux or MacOS, which results in significantly faster runtime performance.
• Adds explicit support for Python 3.7 and 3.8.
• Drops support for Python 2.7 and 3.4.

v1.5.0

• Scaper now returns the generated audio and annotations directly in memory, allowing you to skip any/all file I/O!
• Saving the audio and annotations to disk is still supported, but is now optional.
• While this update modifies the API of several functions, it should still be backwards compatible.

v1.4.0

• Operations on all files happen in-memory now, via new PySox features (build_array) and numpy operations for applying fades.
• Scaper is faster now due to the in-memory changes.