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: Variation in cutoff frequency when using Replay Gain (Read 4958 times) previous topic - next topic
0 Members and 1 Guest are viewing this topic.

Variation in cutoff frequency when using Replay Gain

Hello , I wanted to share with you something that is happening to me to apply to the Replay Gain FLAC files.

I have a file of music in FLAC format obtained from its own original source. Without applying replay gain , cutoff frequency of this is 22Khz . Now I apply replay gain in 89dB , either from the foobar2000 or the command line with flac.exe , 20.5 khz low , but if I compare both spectogramas see that they are identical , they have not changed .

I wonder if I'm losing quality to apply replay gain , which I think not because the replay gain value is stored in the metadata file and notes are not changed , no recoding .

Thanks
jab66


Variation in cutoff frequency when using Replay Gain

Reply #1 –
I can't imagine how that could happen unless you are also downsampling.

There's probably something screwy with your spectrogram software.

If you want to compare spectrograms it would be best to reverse the gain change so you are comparing apples-to-apples.   

It might also be interesting to apply the same gain change that Replay Gain is applying with another tool to see if your spectrogram tool reports the same results.

Variation in cutoff frequency when using Replay Gain

Reply #2 –
The program I use to determine the cutoff frequency is the " spectro " .

No replay gain cutoff frequency is 22.1 kHz.

If I apply rg with foobar2000 v.1.3.8 :
reference loudness: 89 dB
cutoff: 20.5 khz.
track gain: -7.33 dB

If I apply rg with metaflac :
reference loudness: 89 dB
cutoff: 20.2 khz.
track gain: -8.15 dB

You who advise me to do ?

Variation in cutoff frequency when using Replay Gain

Reply #3 –
View the images:

No RG


with RG


I don't know of another program that shows the cutoff frequency. You know another ?

Variation in cutoff frequency when using Replay Gain

Reply #4 –
There's nothing going on here except a less than perfect software algorithm misleading some poor soul who is blindly misplacing his faith in it.

Replaygain adjusts level; that's all.

Besides engaging in a completely academic exercise in futility, there is absolutely no point in this.

Variation in cutoff frequency when using Replay Gain

Reply #6 –
There is no change in "cutoff frequency", regardless of what your software is telling you.  It is only a change in gain.  There is no filter.

Variation in cutoff frequency when using Replay Gain

Reply #7 –
If you're going to try anything, try this...
Quote from: DVDdoug on
It might also be interesting to apply the same gain change that Replay Gain is applying with another tool to see if your spectrogram tool reports the same results.

Variation in cutoff frequency when using Replay Gain

Reply #9 –
With the ReplayGain adjustment of about -8dB you have reduced the high frequencies by 8dB (and everything else by 8dB).   

There's not a lot of high frequency energy to begin with (in any normal audio file) and I suspect you seeing that reduction in level which is more noticeable at high frequencies where there isn't much signal to begin with.

Variation in cutoff frequency when using Replay Gain

Reply #11 –
Looking at the images, it seems to me that the spectrograph software is misinterpreting a gain near -100 dB (which is often translated as digital silence with 16-bit samples) as a "cutoff."

Quote from: jab66 on
I don't know of another program that shows the cutoff frequency. You know another ?

To my knowledge there is no software that can tell you with 100% accuracy whether a cutoff filter has been applied. It can only be inferred from examining the frequencies present in a waveform. In the case of this particular software, it guesses (badly) by examining the gain of frequencies. ReplayGain does not apply a filter.

From the "Feedback" page for the software: "For now please use [popular music piracy site] for feedback". Yeah, definitely dumb software telling you something that isn't true.

Variation in cutoff frequency when using Replay Gain

Reply #12 –
The OP may simply doublecheck with a different spectral view with a freeware like http://www.wavosaur.com/screenshot.php
Is troll-adiposity coming from feederism?
With 24bit music you can listen to silence much louder!

Variation in cutoff frequency when using Replay Gain

Reply #13 –
Or Spek:
http://spek.cc


But the explanation has already been given there:
Quote from: DVDdoug on
With the ReplayGain adjustment of about -8dB you have reduced the high frequencies by 8dB (and everything else by 8dB).

Variation in cutoff frequency when using Replay Gain

Reply #14 –
Yes, it is. I thought the simple spectral comparison will make it easy to see the response is still the same.

Edit: The Spectro software used seems to include RG detection and uses its values for the graphs. There is no explanation on its website only an installer i won't use. It may already include an option to deactivate RG detection.
Is troll-adiposity coming from feederism?
With 24bit music you can listen to silence much louder!

Variation in cutoff frequency when using Replay Gain

Reply #15 –
Very good comments , I've learned some things I did not know until now . The idea of using cutoff was to determine whether the source is lossless when I download a file from the Internet.

Then I believe that the best way is to see the frequency spectrum without cutoff . I do not know if there's another way to detect it.

Variation in cutoff frequency when using Replay Gain

Reply #16 –
The difference in the color spectrum background I went up , it changes only in the Spectro program. I did the same in Audacity and the colors do not change.

Variation in cutoff frequency when using Replay Gain

Reply #17 –
Quote from: Zarggg on
Looking at the images, it seems to me that the spectrograph software is misinterpreting a gain near -100 dB (which is often translated as digital silence with 16-bit samples) as a "cutoff."

Quote from: jab66 on
I don't know of another program that shows the cutoff frequency. You know another ?

To my knowledge there is no software that can tell you with 100% accuracy whether a cutoff filter has been applied. It can only be inferred from examining the frequencies present in a waveform. In the case of this particular software, it guesses (badly) by examining the gain of frequencies. ReplayGain does not apply a filter.

From the "Feedback" page for the software: "For now please use [popular music piracy site] for feedback". Yeah, definitely dumb software telling you something that isn't true.



It can occur because this adds 2 seconds at the end of the file?

Variation in cutoff frequency when using Replay Gain

Reply #18 –
No, the length of the file has nothing to do with this.

To restate: ReplayGain does not apply a cutoff filter. There is no variation in cutoff frequency. The software you used to draw that conclusion is wrong.

Quote from: jab66 on
The idea of using cutoff was to determine whether the source is lossless when I download a file from the Internet.

Then I believe that the best way is to see the frequency spectrum without cutoff . I do not know if there's another way to detect it.

There is no reliable way of determining whether a cutoff filter has been applied without having the source waveform to compare with. Software can guess, often badly, sometimes incorrectly. This is an example of software that guesses incorrectly.

Quote from: jab66 on
The difference in the color spectrum background I went up , it changes only in the Spectro program. I did the same in Audacity and the colors do not change.

The color of the spectrograph indicates the gain (or level) of the frequency in question. Naturally, lowering the gain will cause a difference in the color representing it. The high frequencies already have low levels and ReplayGain is reducing the overall gain of the waveform by a constant amount. Based on the line graph in the screenshots that you posted, it looks like that the software is (erroneously) assuming that frequencies below a certain gain threshold (somewhere slightly above -100 dB) do not exist, which is being interpreted as a "cutoff frequency."

Based on other users posts, it seems that the software you are using is reading the ReplayGain tags and applying that gain change in its analysis. Audacity does not. That's why you see a change in Spectro but not Audacity.

This is a false positive caused by faulty algorithms. Nothing more.

Variation in cutoff frequency when using Replay Gain

Reply #19 –
Quote from: Brazil2 on
But the explanation has already been given there:
Quote from: DVDdoug on
With the ReplayGain adjustment of about -8dB you have reduced the high frequencies by 8dB (and everything else by 8dB).

Agreed.

Variation in cutoff frequency when using Replay Gain

Reply #20 –
Perfect.

I want to use the Tau Analyzer to check the audio, but I do not want to make the recording on a physical disk; I want to generate, from the wav files, an image of CD audio. This is to know if they pass this testing before recording.

Variation in cutoff frequency when using Replay Gain

Reply #21 –
Quote from: jab66 on
Then I believe that the best way is to see the frequency spectrum without cutoff . I do not know if there's another way to detect it.
There is no way, because you don't know which information is missing to begin with. Unless you have both the original lossless source and the lossy encode. Whatever version you currently have is simply the best you have available. If you think it's not the best one available in general, go buy the record, or bug the artist about it, iff you suspect that the producers used lossy encoding in the production chain. Even then you probably cannot do anything, and with the current state of (pop) music production, lossy encoding is the least of your worries w.r.t audio quality.
It's only audiophile if it's inconvenient.