Skip to main content ↑ ↓

Notice

Please note that most of the software linked on this forum is likely to be safe to use. If you are unsure, feel free to ask in the relevant topics, or send a private message to an administrator or moderator. To help curb the problems of false positives, or in the event that you do find actual malware, you can contribute through the article linked here.
Topic: ReplayGain equal loudness filter (Read 10433 times) previous topic - next topic
0 Members and 1 Guest are viewing this topic.

ReplayGain equal loudness filter

Hi

I`d like to ask which "equal loudness filter" is used in f2k ReplayGain is it this one:

Equal Loudness Filter

or is new ISO 226:2003 equal-loudness contour

ISO 226:2003

if old one, maby update?

btw

I did some tone frequency samples to check how ReplayGain equal loudness filter works;
"ReplayGain puts wrong gain on ~4-5,5 kHz, ~7-8 kHz and ~11-12 kHz - don`t have equal volumes comparing to the rest of the test tone freguency samples."

Test files:

ReplayGain.f2k.rgscanner.test

If it`s couse different hardware have different dB/frequency response, than maby there should be ability to adjust volume on individual frequency on ReplayGain playback - reference 1kHz and frequency step generator, it would leave ReplayGain values in files as they are, it would only adjust it`s volume levels to ourselves and hardware we have

thx

ReplayGain equal loudness filter

Reply #1 –
AFAIK everything uses this one...

http://replaygain.hydrogenaudio.org/equal_loudness.html

...though filters have been calculated for many more sample rates than are provided at the very end of that page.


Adopting the new ISO curve may or may not improve performance. The main "problem" with the curve in ReplayGain is that it's an "equal loudness for pure tones" measure - and music rarely consists of pure tones, especially at higher frequencies. As such, it's already a huge approximation, and tweaking it to match the new ISO curve isn't going to solve the fundamental "problem".

A genuine improvement might be gained by amending the curve to match some arbitrary average of equal loudness of pure tones + equal loudness of noise bands over various bandwidths.

Or just redesign the whole of ReplayGain to be based around a cochlea excitation model of perceived loudness, or some approximation to it.

Someone has also suggested using the new standardised loudness measure from the ITU...
http://webs.uvigo.es/servicios/biblioteca/...;!PDF-E.pdf
...but I have my doubts.

I'm also unable to work on improving ReplayGain, but it's a public domain algorithm with open source implementations, so anyone can try.

Cheers,
David.

ReplayGain equal loudness filter

Reply #2 –
It's worth noting that one of the extant cochlear models out there (HEIMDAL) does not perform statistically better than the algorithm used in BS.1770. While the latter (and by extension most forms of weighted RMS loudness estimation) are consistently off by a few db, HEIMDAL has much better accuracy for most samples but incredibly poor accuracy for a few. Unfortunately I forget the exact ref here but I know it's from JAES or a convention.

ReplayGain equal loudness filter

Reply #3 –
Could You tell me:
1. What is the estimated accuracy in achieving it`s reference SPL?
2. For PCM max peak value is 1, since it`s mathematical thing can ther really be a clipping in PCM?
3. Audiacity for PCM track`s that have peak value 1 shows clipping - changing peak level to 1.000016 or 1.000046; Audiacity shows no clipping in that sort of files, my question is - this ir really clipping or mathematical error?
4. Setting "aply gain and prevent clipping according to peak" for file that is quiet but has peak value 1, is that means that file will not be amplified to referance SPL level couse it`s peak value is to high?
5. If PCM audio has peak value 0.9 and gain +10dB will it be amplified 10+ dB or only until it reaches it`s peak level of 1, I`m talking about "aply gain    and prevent clipping according to peak"

Thx

ReplayGain equal loudness filter

Reply #4 –
Though this is an old topic... i wanted to know if anyone has actually worked the iso 226:2003 into replaygain for foobar/ any other app.
I would actually want to try it out if not figure out what is needed to do so.
It seems quite interesting even if it doesn't work well. 
 

ReplayGain equal loudness filter

Reply #5 –
Sorry to bump this, but there's been quite some discussion about ReplayGain (RG) lately, so it seems appropriate to me.

www.speech.kth.se/prod/publications/files/3319.pdf

This is an interesting Master's thesis, a comparative study on file-based loudness estimation algorithms incl. ITU BS.1770 (see David's post) and RG. Here are my main conclusions drawn from the thesis:

  • Exact equal-loudness weighting in the range above ~500 Hz or so seems irrelevant. Some of the best-performing algorithms don't use any weighting above that frequency.
  • RG and BS.1770 both perform very well on average, with BS.1770 producing slightly less errors. IIRC this coincides with the findings reported in this AES 2010 paper.
  • A gating function for ignoring quiet passages in the loudness estimation can improve the performance of BS.1770. See also EBU document 3341 on loudness metering.


Since David posted the original RG equal-loudness filter coefficients before 2003, I assume he used the old ISO curves. Now, what to do? I'm also in favor of adopting the BS.1770 algorithm to compute RG values (it's public domain). The remainder of the RG specification could of course stay the same. Moreover, we could try to improve BS.1770 similar to what has been done in EBU R 128 (esp. "integrated mode" in doc 3341 above) and check whether the "Beatles-Fusion/Rock/Metal" loudness mismatch reported by greynol goes away.

What do you think?

Chris
If I don't reply to your reply, it means I agree with you.

ReplayGain equal loudness filter

Reply #6 –
Quote from: C.R.Helmrich on
What do you think?

This is where I would like to see RG headed. The ITU and EBU work has been well tested is well maintained and is becoming widely used. Pointing RG at this work would apparently improve performance and would definitely reduce maintenance and improve interoperability.

ReplayGain equal loudness filter

Reply #7 –
Quote from: Notat on
Quote from: C.R.Helmrich on
What do you think?

This is where I would like to see RG headed. The ITU and EBU work has been well tested is well maintained and is becoming widely used. Pointing RG at this work would apparently improve performance and would definitely reduce maintenance and improve interoperability.


I fully subscribe these words.
ITU and EBU are nowadays the standards for measuring loudness and therefore the new EBU algorithm should be integrated in RG instead of the algorithm of 2001.
This will provide a kind of future for RG in terms of professionalism.

Kind regards

Jean

ReplayGain equal loudness filter

Reply #8 –
Two AES papers on the gating feature described in the above EBU document 3341:


Conclusion: the relative gating configuration in EBU R 128 is well chosen, although the -8 dB threshold might be a bit on the high side (-9 to -11 dB might give a tiny performance improvement on average, according to the second paper).

Chris
If I don't reply to your reply, it means I agree with you.

ReplayGain equal loudness filter

Reply #9 –
-8db? Interesting. I guess I need to look closely at those papers sometime; when I was writing pfpf in 2008 I (quite subjectively) figured that the best gate thresholds were in the -20 to -30db range. But I think I may have been using substantially different window lengths.