One of the great things about blogs is that it allows scientists to express their ideas in a more open and direct way. The old way to criticize a paper was to write a letter or a paper and try to get it published. Now, anyone can provide arguments and discuss a paper for the world to see.
In 2004 I tried to publish a rebuttal of the paper “Potential causes for amphibian declines in Puerto Rico” by Burrowes, Joglar and Green, published in Herpetologica. Unfortunately, the editors decided that the wording in their paper was vague enough that it made it hard for them to justify my comments as a rebuttal. I recently stumbled upon the original emails, so I decided to make my original comments of that paper available here. As a disclaimer, I worked with the first two authors for several years and I was one of the students that collected the data used in that paper during that time.
The full citation of the paper is:
Burrowes, Patricia A., Rafael L. Joglar, and David E. Green. 2004. Potential causes for amphibian declines in Puerto Rico. Herpetologica 60: 141-154.
Some of the problems of the paper are: Continue reading
I’ll be attending the SACNAS National Meeting this weekend in Dallas, TX. If you’re there, just look for me.
A few days ago the paper appeared in the online list of Ecological Informatics:
Abstract below the fold.
High quality audio files take a lot of space, for example, a 1 minute stereo sound at CD-quality (44.1 kHz) will take about 10-11 Mb. A large collection of sound files will fill hard drives really fast. The usual solution to this problem is to compress the files, however each compression system has its limitations.
The common zip compression format works well for text but does not compress audio in any significant degree. Encoding the file to mp3 will destroy a lot of information that will be necessary to analyze the audio.
FLAC (Free Lossless Audio Codec) is optimized for wav audio files, and can compress a wav to 30-50% of its original size. It appears that the amount of compression will depend on the complexity of the sound recorded. The decoding back to wav is very fast and it is an exact copy of the original file. Continue reading
I am posting the paper and the data associated with this paper:
Villanueva-Rivera, L. J. 2007. Digital recorders increase detection of Eleutherodactylus frogs. Herpetological Review 38: 59-63. PDF.
Read the readme.txt file in each zip archive for details.
This script uses the numpy and audiolab modules to generate waveform and spectrogram png images from a wav file. The original script was from the Freesound Blog. The wav file needs to be 16 bit, mono and have a sampling rate of 44.1 kHz. The script takes about 2-3 seconds for a 60 seconds file (on a Core2 Duo 2GHz, 2GB RAM machine).
Update: The script is now a project on SourceForge: http://sourceforge.net/projects/soundviewer/