"en.wikipedia.org/wiki/EBU_R_128"
"tech.ebu.ch/loudness"
"tech.ebu.ch/files/live/sites/tech/files/shared/r/r128.pdf"
"tech.ebu.ch/files/live/sites/tech/files/shared/tech/tech3341.pdf"
"tech.ebu.ch/files/live/sites/tech/files/shared/tech/tech3342.pdf"
"tech.ebu.ch/files/live/sites/tech/files/shared/tech/tech3343.pdf"
"tech.ebu.ch/files/live/sites/tech/files/shared/tech/tech3344.pdf"
"itu.int/dms_pubrec/itu-r/rec/bs/R-REC-BS.1770-5-202311-I!!PDF-E.pdf"
There should be some set of test files available, but I didn't
succeed to download it.
SoX already has a "loudness" feature but it does not seem to be the
one defined by "EBU R 128" & "ITU-R BS.1770". It also has "stat" and
"stats" features, with findings not including values used by
"EBU R 128" & "ITU-R BS.1770"
Suggestion: add evalutaion of those 4 values
* true peak
* max Momentary loudness (400 ms)
* max Short term loudness (3s)
* Integrated loudness (whole file)
and make them available in a suitable way ("-stat" or "-stats" or a new
command).
Indeed I pefer having those values reported, over performing some
"normalization" directly. Having those values I can perform the task
either with SoX or other program (AudaCity, FFMpeg, ...). Plus
the "gain" feature of SoX is already overcomplicated.
Maybe the options "norm" and "gain n" should be tagged as discouraged
in the manual, since they seem to perform the highly
unfortunate "peak normalization" (true or untrue?).
I have found some "loudness" in SoX, AudaCity, and FFMpeg. Unfortunately
SoX seems to use a different loudness than "EBU R 128" & "ITU-R BS.1770".
AudaCity apparently can normalize correctly, but not report any findings,
plus it's bloated and GUI-only. FFMpeg mentions "EBU R 128" but seems
to do something strange, or different from what AudaCity does at least.
_______________________________________________
SoX-devel mailing list
SoX-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/sox-devel
