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 (Read 7161 times) previous topic - next topic
0 Members and 1 Guest are viewing this topic.

Replaygain

So I was fooling around with replaygain, but I notice when comparing older tracks with newer more modern tracks that the newer tracks sound cruddier. I mean the newer tracks have blaring deep bass and the volume drop seems to cut out the mids and highs causing the bass to equal the old tracks. The point of Replaygain is to make all tracks sound equal, but when comparing the older tracks they sound bass shy while the newer tracks have too much bass and cut the mids and highs. So my question, is there a non-destructive equalizer similar to Replaygain what will tame those bass frequencies to truly balance out with the older tracks?

Replaygain

Reply #1
The point of Replaygain is to make all tracks sound equal, but when comparing the older tracks they sound bass shy while the newer tracks have too much bass and cut the mids and highs.


People tend to perceive a change in amplitude as a change of bass.  That is, lowering the volume will cause lower frequencies to become less noticeable relative to higher frequencies.

So my question, is there a non-destructive equalizer similar to Replaygain what will tame those bass frequencies to truly balance out with the older tracks?


Thats what a normal EQ does. 

Replaygain

Reply #2
People tend to perceive a change in amplitude as a change of bass.  That is, lowering the volume will cause lower frequencies to become less noticeable relative to higher frequencies.



Thats what a normal EQ does.


Actually, my problem is lowering the volume the lower frequencies become more noticeable and higher ones become less. I would like to set the eq to a specific album and not have to worry about a manual one.

Replaygain

Reply #3
theoretically you could re-run replaygain scanner after the high-pass filter of your choice.
PANIC: CPU 1: Cache Error (unrecoverable - dcache data) Eframe = 0x90000000208cf3b8
NOTICE - cpu 0 didn't dump TLB, may be hung

Replaygain

Reply #4
Quote
The point of Replaygain is to make all tracks sound equal
The point is to make the volume equal without otherwise muking with the sound.

Quote
So my question, is there a non-destructive equalizer similar to Replaygain what will tame those bass frequencies to truly balance out with the older tracks?
Not exactly...  If you apply an equalizer at playback time, of course it's non-destructive.  But I don't know of an equalizer that will apply automatically apply different EQ to different files. 

I also don't believe it's possible to EQ automatically the way ReplayGain or MP3Gain adjusts the volume automatically without human intervention.  There are some "EQ matching" tools, but I seriously doubt they're useful over a complete music library, and I doubt they are useful at all without using some human judgement...

Equalization of a WAV file is essentially lossless (as long as you don't drive the volume into clipping), but in general it's not  perfectly reversible.  (Of course you can save a copy of the unaltered original.)

Since this is the MP3 forum... As far as I know there are not any lossless MP3 equalizers...  If you want to equalize an MP3, it has to be decompressed, EQ'd, and then re-compressed (going through a 2nd lossy compression step).    It might be possible  to EQ an MP3 without decompressing first by manipulating the FFT information, but I've never heard of that being done.


Replaygain

Reply #6
Since this is the MP3 forum... As far as I know there are not any lossless MP3 equalizers...  If you want to equalize an MP3, it has to be decompressed, EQ'd, and then re-compressed (going through a 2nd lossy compression step).    It might be possible  to EQ an MP3 without decompressing first by manipulating the FFT information, but I've never heard of that being done.


The MP3 decoder of Winamp ( 1.x and 2.x days IIRC. ) used to equalize the files by modifying the gains of the scalefactors. That's why, originally, the equalizer only worked with MP3 files and not with the other input plugins (wavs, mods,...).
Later on, they added a generic equalizer and there was an option to switch between the new and the old equalizer for MP3 files.
Just checked the version of 2.95 that I have, and still had it: properties of the mpeg audio plugin, decoder tab "Fast layer 3 EQ".

Of course, storing this back to the file would be problematic, being that the gains are in 1.5dB steps.

Replaygain

Reply #7

Since this is the MP3 forum... As far as I know there are not any lossless MP3 equalizers...  If you want to equalize an MP3, it has to be decompressed, EQ'd, and then re-compressed (going through a 2nd lossy compression step).    It might be possible  to EQ an MP3 without decompressing first by manipulating the FFT information, but I've never heard of that being done.


The MP3 decoder of Winamp ( 1.x and 2.x days IIRC. ) used to equalize the files by modifying the gains of the scalefactors. That's why, originally, the equalizer only worked with MP3 files and not with the other input plugins (wavs, mods,...).
Later on, they added a generic equalizer and there was an option to switch between the new and the old equalizer for MP3 files.
Just checked the version of 2.95 that I have, and still had it: properties of the mpeg audio plugin, decoder tab "Fast layer 3 EQ".

Of course, storing this back to the file would be problematic, being that the gains are in 1.5dB steps.


That's actually what MP3Gain does. It modifies the scale factor in each frame of the MP3 file, making a lossless change to the file. Usually, replaygain is added on top of this for EQing finer than the 1.5dB granularity you mention.

Replaygain

Reply #8
Well, no.  MP3Gain performs amplification by adjusting the global gain field in each frame.  This is different than adjusting each of the individual scalefactor bands within each frame in order to apply equalization.

Replaygain

Reply #9
It applies a global gain change by changing the scale factor bands equally, but I see no reason why this couldn't be used for EQ as well.

Replaygain

Reply #10
changing the scale factor bands equally

This contradicts everything I have read on this forum over the past decade.  I'm not saying you're wrong as I haven't looked at the code.  Have you?

I do think taking equal loudness contours into account when adjusting a track/album's volume to match some reference* would be pretty slick, if it could be pulled off (considering the 1.5dB steps), though.  However, this still doesn't address the stylistic choice of boosting the bass that so often goes along with applying heavy DRC.

(*) then again, you can't exactly assume that all titles were auditioned at the same volume when they were being mastered.

Replaygain

Reply #11
<br />It applies a global gain change by changing the scale factor bands equally, but I see no reason why this couldn't be used for EQ as well.<br />
<br /><br /><br />

because that would mean writing an extra value different for each sample to reshape the wave. while replay gain is but 1 single identical value applied to all sample numbers. it would make more sense to just reencode the all signal. instead your proposition would effectively create a second signal with an independent value per sample. that would make for a bigger file, more work, and no actual benefit over using EQ.

Replaygain

Reply #12
I think there are several confusions after my post.

First off, the knowledge so far is that MP3Gain changes only the global gain.

MP3 has 13(?) or 22 scalefactor bands depending on the frequency used (mpeg 2 or mpeg 1 respectively).
All except the last band have an individual gain, and then there is a global gain that applies to all bands (including the last one, which has no individual gain).
The absense of the individual band gain for band 21 (the index is zero based, so 0 to 21, instead of 1 to 22) is what causes the bitrate increase of some files, that the LAME -Y switch tries to reduce (and which is set by default in all except for the higher VBR modes).

Of course, a band gain cannot decrease indefinitely. IIRC they are stored in 127 possible values (maybe it is less than that. I couldn't find the value by briefly looking on internet). If it decreases to 0, it cannot decrease more.

Also, when I said that this was not very adequate for Equalizing, I meant that it could represent an error between bands of +1.5dB, and equalization usually requires quite subtle modifications sometimes. (plus, the added problem of the missing scalefactor band gain)


@casteofargh : I don't really understand what you are saying. MP3 does not encode in time domain, it encodes in the frequency domain. (MDCT). And as I've said, the stored bands already have individual gains.

Replaygain

Reply #13


@casteofargh : I don't really understand what you are saying. MP3 does not encode in time domain, it encodes in the frequency domain. (MDCT). And as I've said, the stored bands already have individual gains.


but does replay gain have anything to do with the mp3 encoding? how I imagine things, the mp3 get extracted to PCM for the DAC to use, and the replay gain value is simply added in the process to make a loudness difference by changing each PCM sample by the same amount effectively changing the asked voltage over the entire track.  it requires almost no processing, needs the gain to be stored only once per track, so that's how I imagine they implemented it.
am I wrong about that educated guess?

Replaygain

Reply #14
You are not wrong, but you missed the point that we are not talking only about replaygain, but a concrete implementation named MP3Gain.

MP3Gain uses the replaygain (1.0) algorithm to determine a track and album gain, and then the user chooses to apply one of them to the file, effectively modifying it. In other words, with MP3Gain, the player does not need to know about replaygain, because the file is modified.

This modification is done by altering one value of each mp3 chunk of the file which is called "global gain".
The Mp3 format uses this global gain (and the other scalefactor band gains) as a means of reducing the precision of the data that it has to store. (to put it simple, if there are less discrete values to store, it can be stored with less bits, and then the gain compensates for the difference).

So MP3Gain makes use of this existing feature, for something slightly different to what it was meant.

Replaygain

Reply #15
Nobody is talking about mp3gain.

I guess my short post was ignored, so let me try again:

calc faze:
mp3 > (inverted) eq of your choice (high.pass of some sort) > replaygain calculus > storage

play faze:
mp3 > player that can take "storage" and apply it > playback

or use mp3gain somehow to burn that external storage into mp3.
PANIC: CPU 1: Cache Error (unrecoverable - dcache data) Eframe = 0x90000000208cf3b8
NOTICE - cpu 0 didn't dump TLB, may be hung

Replaygain

Reply #16
How will that adjust tracks so that they have the same the tonal balance?  The OP wants to "tame the bass frequencies" of newer tracks.  Having these newer tracks play louder isn't going to achieve that. According to Fletcher-Munson they'll sound even more bass heavy.

Replaygain

Reply #17
How will that adjust tracks so that they have the same the tonal balance?  The OP wants to "tame the bass frequencies" of newer tracks.  Having these newer tracks play louder isn't going to achieve that. According to Fletcher-Munson they'll sound even more bass heavy.



He wants something like Audysseys' 'Dynamic EQ'.


http://www.audyssey.com/technologies/dynamic-eq


Replaygain

Reply #19
How will that adjust tracks so that they have the same the tonal balance?  The OP wants to "tame the bass frequencies" of newer tracks.  Having these newer tracks play louder isn't going to achieve that. According to Fletcher-Munson they'll sound even more bass heavy.

That would require a subjective test using op/s technology and ears, which is free to perform for himself, assuming that middle range will be kinda equal in loudness, but can't exactly claim that (and not going to test either, anyway if it fails, reverse eq to positive and retry).
PANIC: CPU 1: Cache Error (unrecoverable - dcache data) Eframe = 0x90000000208cf3b8
NOTICE - cpu 0 didn't dump TLB, may be hung