Release Notes

Version 2.1

We encourage your feedback, especially if you notice any bugs or have suggestions for improvements and new features. Please contact us at songscope@wildlifeacoustics.com or call toll-free +1 (888) 733-0200.

System Requirements

All platforms require an Internet Connection during installation (for product activation after purchase). If a sound card is present, Song Scope can play audio portions of recordings.
Windows 95/98/2000/NT/XP/Vista
Linux RedHat i386
Mac OS X 10.4 universal binary (ppc/i386)

Limitations

Only PCM (uncompressed) .WAV files (RIFF/WAVE) and Audio IFF files are supported in addition to the Wildlife Acoustics Audio Compression (WAC) format.

Version 2.1A

Bug fix: Audio playback on Microsoft Vista

Another bug causing hangs was found in the Vista implementation of a Microsoft system library call (in particular waveOutGetPosition()). We have worked around the problem by avoiding the problematic call.

Version 2.1

Bug fix: Audio playback on Microsoft Vista

There was a bug in which audio playback did not work under Vista causing Song Scope to hang. This has now been corrected.

wxWidgets upgrade

Upgraded to wxWidgets 2.8.8 library from 2.8.7. This should fix a number of bugs and improve performance of the graphical user interface.

Version 2.0: Changes from Release 1.12

Compatibility

Version 2.0 can read .ssr recognizer files created from older versions of Song Scope. However, recognizer files created in version 2.0 can not be used by older versions.

Algorithm Improvements

In version 2.0, there have been several improvements made to the classification algorithms. By default, recognizers built in version 2.0 will use the new 2.0 algorithms. However, if you prefer to build recognizers using the old 1.0 version, you can choose 1.0 in the new Algorithm control on the Detector control panel.

To upgrade .ssr recognizer files from older versions, open the file, change the Algorithm control on the detector control panel to 2.0, then regenerate and save the new recognizer. You may want experiment a little for best results. In particular, the new algorithms seem to work best with a smaller Max Resolution value in the 4-6 range. You may also be able to increase the Dynamic Range a little for better results.

Quality score improvements

In version 2.0, quality values are now in the range 0-100. A quality value of 100 indicates that the duration (time), density (symbols), and complexity (states) of a vocalization are average compared to the training data. Lower quality values indicate a greater distance from the mean compared to the training data. A minimum quality value of 50 should allow approximately half of the training data through, and a minimum quality value of 10 should allow approximately 90% of the training data through. Quality values below 10 are outliers that are less likely to be a positive match.

History

Version 2.0

Bug fix: Windows audio playback

On some Windows platforms, playback sound was choppy. This has been fixed with improved buffering.

Previous Page Next Page