paule 0.4.8

What's Changed

• fix for initilizing tensors on the right device by @AndreDerCoole in #4
• pytorch load warning fix by @AndreDerCoole in #6
• nummpy 2.0 deprecation np.Inf fix by @AndreDerCoole in #7

New Contributors

• @AndreDerCoole made their first contribution in #4

Full Changelog: v0.4.5...v0.4.8

v0.4.5

• allows to pass a string to a results file to the visualize.visualize_results function
• updates README

Full Changelog: v0.4.4...v0.4.5

paule 0.4.4

What's Changed

• small fix to initialize a tensor on the target device and not on the cpu by @derNarr in #1
• Speech classifier by @derNarr in #2
• Five point stencil by @derNarr in #3

Full Changelog: v0.3.6...v0.4.4

Adds smiling functionality

• adds the functionality to keep the VTL in a "smiling" state during planning.

paule 0.3.5

• fixes pyproject.toml to work with new poetry version and to allow for newer version ins dependencies

paule 0.3.4

• fixes somatosensory pathway
• fixes calculation of somatosensory representation
• doubles velocity and jerk constraints in somatosensory pathway
• adds continue learning for the somatosensory-to-acoustic model
• fixes path to VTL library on Windows

paule 0.3.1

Still a version in alpha status, but this release adds:

• function to download pretrained weights
• new VocalTractLab version for Linux, MacOS, and Windows
• adds first tests
• adds poetry for maintaining the package
• improves documentation

paule 0.2.0

Major improvements over first release, but still lacking and instruction on how to download the pre-trained weights.

paule v0.1.1

First release with some useful code, but lacking some of the big pretrained model files and the data to directly run it after installation.