Input files are resampled to 44100Hz before DR computation, although this is not mentioned in the linked PDF

[magicgoose]

dr14_t.meter/dr14tmeter/audio_file_reader.py

Line 160 in c29b2ff

return " -i \"%s\" -b:a 16 -ar 44100 -y \"%s\" -loglevel quiet " % (file_name, tmp_file)

— these parameters are used to create temporary files for actual DR measurement, if I understood correctly. And they instruct ffmpeg to resample everything to 44100Hz.

http://www.dynamicrange.de/sites/default/files/Measuring DR ENv3.pdf — here's no mention of that.

What is the reason behind this? If it comes from a reverse engineering of official closed-source tools, then maybe you have some other interesting findings to share? 😄

[magicgoose]

I just tested it on some 96khz record (FLAC variant of http://www.opengoldbergvariations.org/) with and without the resampling, and it seems that resampling makes it closer to the "official" numbers from foobar2000 plugin. 🤔