v0.7.0

This release marks a significant change to the OpenSoundscape API, the addition of several new features, and the resolution of various bugs. Updates include:

• simplified training and prediction with CNNs / pytorch machine learning models
• incorporated preprocessing into CNN class so that all necessary parameters and settings are bundled with saved object
• extract audio from AudioMoth files based on a real-world timestamp
• more flexible spectrogram generation parameters in Spectrogram.from_audio
• load specific channels of multi-channel audio file
• more flexible validation step and choice of metrics when training PyTorch models
• refreshed tutorials

as well as fixes to various bugs and resolution of various issues.

Merged branches

• Tutorial updates by @jatinkhilnani in #475
• add function to load separate channels from audio file by @sammlapp in #492
• Issue 401 change mp3s in tutorials to wav by @louisfh in #490
• Refactor preprocessing and models.cnn by @sammlapp in #489
• v0.7.0 minor release by @sammlapp in #499

New Contributors

• @jatinkhilnani made their first contribution in #475

Full Changelog: v0.6.2...v0.7.0