0.3.0a0

This is the first alpha release of the 0.3 series.
One new feature:

• export log file of sensor fault #46

0.2.5

This is the fifth bugfix release of the 0.2 series.

One bugfix:

• ModuleNotFoundError: No module named 'numba.decorators' #47

0.2.4

This version is functionally identical to 0.2.3

0.2.3

This is the third bugfix release of the 0.2 series.

Two bugfixes:

• NameError: name 'segment_start_str' is not defined (#40)
• Invalid number of frames in input PCEN. Expected 104 but got 0. (#42)

0.2.2

This is the second bugfix release of the 0.2 series.

Two bugfixes:

• local variable 'has_sensor_fault' referenced before assignment (#34)
• ValueError: Output array is the wrong shape (#37)

0.2.1

This is the first bugfix release of the 0.2 series

The only change with respect to 0.2.0 is a bugfix of the HDF5 routine for loading Keras models, which allows the same distribution birdvoxdetect to be run over multiple threads at once.

0.2.0

This is the first stable release of the 0.2 series
Main change with respect to the beta version is the addition of a context-adaptive neural network model.
For more details on this model, read:

Robust Sound Event Detection in Bioacoustic Sensor Networks
Vincent Lostanlen, Justin Salamon, Andrew Farnsworth, Steve Kelling, and Juan Pablo Bello
PLoS ONE 14(10): e0214168, 2019. DOI: https://doi.org/10.1371/journal.pone.0214168

0.2.0b1

This is the first beta release of the 0.2 series.
Main changes with respect to the alpha release:

• threshold now corresponds to positive predictive value (aka precision) on BirdVox-300h. This correspondance is valid in the range 5-65.
• fixed a bug in birdvoxactivate

0.2.0a2

This is the third alpha release of the 0.2 series.
Main changes with respect to the previous alpha release:

• the new default threshold is 50, not 30
• the threshold can now go below 10
• the top confidence is now 99.99, not 100. This standardizes the char length of the exported clip names
• the last 150ms of the last chunk have been removed from the detection to avoid a BVD size mismatch error
• the values of the detection curve needs have been tripled to cover the range 0-99 better

0.2.0a1

This is the second alpha version of the birdvoxdetect v0.2 series
Compared to the previous alpha version, the h5py dependency has been relaxed to >=2.7.0
This circumvents a pip install errror