Skip to main content
Topic: weighted curves in audio... (Read 1837 times) previous topic - next topic
0 Members and 1 Guest are viewing this topic.

weighted curves in audio...

I was looking this up on wikipedia and was wondering if the ITU-R 468-weighting (combined with a C-weighted curve) is better than an A-weighted (C-weighted curve) for loudness estimation. The article had that the ITU-R 468 was tested differently than the A-weighting. (A-weighting was tested with pure tones.)
I think the replaygain site said that it used a C+A weighted curve... 
It seems pretty interesting to see the difference between the two...

weighted curves in audio...

Reply #1
I was looking this up on wikipedia and was wondering if the ITU-R 468-weighting (combined with a C-weighted curve) is better than an A-weighted (C-weighted curve) for loudness estimation. The article had that the ITU-R 468 was tested differently than the A-weighting. (A-weighting was tested with pure tones.)
I think the replaygain site said that it used a C+A weighted curve...  :huh:
It seems pretty interesting to see the difference between the two...

Briefly, loudness estimation of what?  ITU-R 468 is an appropriate, indeed probably the best, weighting for measuring low level noise, that is, noise that is much quieter than the desired audio.  For that purpose in fact, A-weighting, although widely used, does not correlate with subjective impression as well.  On the other hand, if you are interested in the loudness of the desired audio at intermediate levels, say speech, something a few tens of dB louder than the background noise, then A-weighting gives a reasonable answer.  For still louder things, jet engines or rock concerts, still other weighting curves will correlate with the pain more closely.

Ken G, San Francisco

weighted curves in audio...

Reply #2
Are there any "unified" attempts?

It would be a lot easier to just have a single reading from the calibrated measuring device that applied a good linear weighting as a function of SPL.

-k

 

weighted curves in audio...

Reply #3
Are there any "unified" attempts?

It would be a lot easier to just have a single reading from the calibrated measuring device that applied a good linear weighting as a function of SPL.

-k


Well, alas truth is not easy!  Yes, of course there are other methods of measuring subjective loudness, as you put it, "unified" methods, but they are much more complex, requiring splitting the audio into narrow frequency bands, measuring the level in each, subjecting those levels to non-linear adjustments because of the non-linearity of human hearing and masking effects, and finally summing the result to a single figure.  No simple weighting filter will give good correlation with subjective loudness over a wide range of levels and frequencies.  Look at the Fletcher-Munson equal loudness curves; they are not equally spaced, or putting in another way, they have different shapes at different levels.  Furthermore they are for signals at one frequency at a time, and therefore do not show masking effects.

Ken G, San Francisco

 
SimplePortal 1.0.0 RC1 © 2008-2020