DJing Discussion

This area is for discussion about DJing in general. Please remember the community rules when posting and try to be polite and inclusive.

MP3Gain Sound Problems

DJJUSTICE 10:59 PM - 22 December, 2006
Ok So ive got the killer library now, serato is very stable, finding good legal 320K sources of music DAILY (see record labels? we dont mind paying) and all is well.

BUT

A common complaint from Club owners is that even though I used MP3Gain to gain all my files to 92db some songs "sound" louder than others. The owner came to me and said "dude there are like drastic volume changes" I told him he was crazy because i can see the peak level meters at a clean 0 on the mixing board, everything is set to unity at 7. PLUS I have mixer gain control per channel before I even mix it in. It should be perfect. But this was not the case as he had me come out to the floor and said "see?" ..

Now as best I can explain it is this. Some songs may have the "Meat" of the song averaging 92db and others have peaks averaging 92db so they are ultimately lower. even though the peaks are the same. It may be the same phenomenon when I open up a record label track in cool edit and notice how "big" the waves are. Its like everything is real close to the peak line. Other songs everything is in the middle and only certain peaks are close to the peak line. Is this some sort of compressor? Well I may be going on a tangent but Im trying to have MP3 gain make all files the same.. Like vinyl.. Should I use thier NOT RECOMMENDED setting of "making mp3's as loud as possible without clipping"? .. I think right now I have it set to 92db but dont clip tracks.

Its a real problem and I can especially SEE it when looking at a recording after the fact. You can clearly see the drop in DB even though they "are all the same volume"..

I hope I made any sense. Please help
djaction 11:13 PM - 22 December, 2006
interesting.. I have it set to gain volume even if it clips and everything set @ 94db. I *thought* that all my songs would be look the same or be very close when cue'd up and playing through my mixer's meteres. They don't.. some definite differences and I'm guessing your right about the averages and peaks.

Would be nice if there was a way to have the DB set for the entire track.
DeezNotes 11:23 PM - 22 December, 2006
Quote:
Would be nice if there was a way to have the DB set for the entire track.


I *thought* that's how it worked, but you shouldn't take my word for it. I usually skim over those details when I read.

Justice, it sounds like your problem is the "loud as possible without clipping" option. If you set everything to 92 and a track clips at 89, that track will be set to a lower value so it will not clip. Me personally, I allow for clippage to get all my tracks the same volume (95db). I do, however set my instrumentals and acapellas to 93db as they have an overall lower volume than the full track. If you set all 3 versions (vocal, inst and acap) to the same volume (and allow clipping) you'll see what I mean.

My suggestion is to play around with a handfull of tracks and the options MP3gain sets. You can change the volume in MP3gain while SSL is open... just don't have the track loaded.

Let us know what you come up with.
DJJUSTICE 11:40 PM - 22 December, 2006
I also just found this article

en.wikipedia.org

Definately a mouthfull, I will play with serato and MP3Gain but I still need to know the ultimate setting for MP3Gain.
DJJUSTICE 11:56 PM - 22 December, 2006
Quote:
I allow for clippage to get all my tracks the same volume (95db)


DeezNotes.. Do you have any problems with complaints of volume changes? Also i notices this phenomenon is not present on my mixed cd's for whatever reason. (ok I think its because Im looking at the outputed DB levels live in the recording after the mixer so I adjust it in real time so everything is level??)
djaction 12:12 AM - 23 December, 2006
Quote:
Quote:
I allow for clippage to get all my tracks the same volume (95db)


DeezNotes.. Do you have any problems with complaints of volume changes? Also i notices this phenomenon is not present on my mixed cd's for whatever reason. (ok I think its because Im looking at the outputed DB levels live in the recording after the mixer so I adjust it in real time so everything is level??)


this is kind of a shot in the dark but something I didn't think about before. If the problem doesn't occur with mix cd's, maybe its a problem with the mixer or amps in the club?
DJJUSTICE 12:28 AM - 23 December, 2006
No i think its definately a problem with peak vs average on mp3 gain and the tracks themselves. The probem was worse on a UREI with no db meters. When I used my pioneer djm600 I could see the levels. I think when im making a CD I can see the outputed volume on Cool-Edit and use a seperate recording level gain (dj booth output on mixer) to get it as close to 0 as possible during recording, this luxury I dont have in the club
Dj KaGeN 12:38 AM - 23 December, 2006
interesting.... When in a club setting - I would look at the waveform and decide if it had the MEAT (so-to-speak) to be loud and hold the dancefloor's attention, knowing in the back of my mind that all the tracks are technically the same volume. I may have to revisit this, wow - what a task to redo the whole library now.
nik39 1:03 AM - 23 December, 2006
Quote:
A common complaint from Club owners is that even though I used MP3Gain to gain all my files to 92db some songs "sound" louder than others.

Unfortunately... as you already figured, mp3gain is not perfect :( I would say it has a failure rate of maybe 5%

Quote:
Would be nice if there was a way to have the DB set for the entire track.

Deez is correct, thats how it already works.

Quote:
Justice, it sounds like your problem is the "loud as possible without clipping" option. If you set everything to 92 and a track clips at 89, that track will be set to a lower value so it will not clip. Me personally, I allow for clippage to get all my tracks the same volume (95db).

Thats a good approach (I do it the same way with the same reasoning :D ).

Justice, did you check if you played around with the clipping setting?
MSF 4:51 AM - 23 December, 2006
At first I was using mp3gain on my library, but then I re-ripped all my tracks and I don't use Mp3gain anymore.

Personally, I felt that with some bass-heavy tracks, mp3gain would squash the music and lower the gain of track by like -4db or so. This was done to tracks that were ripped from original CD's.

When I had compared the mp3gain'ed track to the original rip of the song, I definetely preferred the track with no mp3gain.... The original seemed to hit harder, obviously because the 'gain' of the file was not altered.... Thats just me though.. some of you may have different preferences with it. (all-subjective)

Nik is right.. mp3gain is definetely not 100% perfect.
DJJUSTICE 9:19 PM - 23 December, 2006
Well I just ran some minor test. With xmas and all I dont have all the time to do a proper scientific test as to why one song sounds louder than the other.

But I did notice that newer songs did have that highly compressed punch as described in the wiki article above. Im talking about house. I need to run some proper test but not really sure as to what to test. I think whats needed is a tool that can truly compare peak vs average db on alot of songs. Looking at them one by one is not really practical, and aside from a real-time compressor, may be the only solution IMO.

If anyone knows about this problem please speak.
nik39 9:27 PM - 23 December, 2006
Quote:
Nik is right.. mp3gain is definetely not 100% perfect.

No automated solution will do a perfect job on doing a "perceived volume" comparison, they will all base their results on their own psycho accoustic model. Even if that was perfect...as humans are not the same, we will disagree on the volumes of certain songs.
DJ A-NAK 12:24 AM - 24 December, 2006
yea just use mp3 gain as a "GUIDE" and do most of the volume and gain changes by ear while Cueing...that's a part of DJ'in anyways...all digital stuff now a days just like Serato are there to "HELP" not do it all for you or that shit would be producin DJ's left and right
dtism 2:28 PM - 18 January, 2007
Hope you don't think I'm trying to hijack your thread DJJUSTICE but, I'm also struggling with getting my levels of MP3's how I want them. I would prefer not to have to use gain adjustment in SSL and just control the gain on the mixer (as if I were playing vinyl or cd). Therefore I would like all my MP3's normalised to be as loud as possible without clipping, but popular opinion seems to suggest that this is a bad thing. I understand the idea that if a track has a loud pop, then the rest of the track will seem very quiet if I normalise this way (however I usually try ant clean pops and clicks from my vinyl rips)

I have tried using mp3gain to set all of my tracks to the same level, but they seem quite a lot quieter than say if I was playing a cd.

I might experiment with 'maximum no clip gain' vs 'track gain' and see what works best for me.
derby 2:42 PM - 18 January, 2007
what level do you have mp3gain set for?
dtism 3:13 PM - 18 January, 2007
I have tried several, at the moment I'm using 92.0
tig ol' bitties 3:39 PM - 18 January, 2007
if you watch what it does, sometimes it wont put it exactly on what you want. So say you have it set at 92, some songs will read in at 92.3 or 91.8, so those are the ones that would have the problem.
I def have noticed this on mine. I set mine at 95 and its as uniformed as I have seen them. 92 to me is too low. Leaves to much room for error.
dj skraps 4:05 PM - 18 January, 2007
tig, so when you st @ 95...i'm sure you get some that say they are clipping....do you normally leave em that way or ajust those differently....i have all mine set at 94 and there are quiet a bit that show clipping, i havent played em out to see for myself...but in your experience would you change those or leave them being that mp3gain is not perfect??
dj skraps 4:06 PM - 18 January, 2007
and what do you mean room for error??
tig ol' bitties 4:10 PM - 18 January, 2007
Not too many. Yes there are some and the ones I notice I do knock them down to 94, but for the most part the ones at 95 are gravy. If they are like 95.3 I will def knock em down, but anything 95.0 or below I will leave.

I would say from my experience out of 100 tracks, 40% of them wont be exact. As in 95.0 perfect. Either be slightly higher or slightly lower. Like I said if its 95.3 or higher I will fix.
tig ol' bitties 4:13 PM - 18 January, 2007
Quote:
and what do you mean room for error??


I dont know, like from what I have noticed when I first started using it (mp3gain) I was using 92 and 93.

Doing the same basic thing, like using 92 or 93 out of 100 tracks I found more like 50-60% of them were either slightly higher or lower compared to 95 where id say its closer to 40% accurate.

This is just my experience with it, results obviously may vary, but I never hear radical volume changes in my mixes using 95 and/or clipping.
derby 4:17 PM - 18 January, 2007
i doubt many can tell a .5dB difference between two tracks, let alone with DJ ears (loud headphones, loud booth monitor, nightclub speakers, etc).

in regards to the above: autogain/mp3gain should not be used as a replacement for using your ears. neither should channel level indicators. it's all about perceived loudness. as mentioned above, a bass-heavy track at 92dB would likely sound quieter than a mid or high-heavy track at 92dB.
tig ol' bitties 4:19 PM - 18 January, 2007
For certain track a .3 difference is noticeable, especially if your just fucking around on little speakers.

I am constintely fucking with my EQs anyway, so I manually keep the sound consistent anyway. I'd say thats the best way to do it, but obviously some tracks def require normalization.

use whatever is clever.
That Guy 1 10:36 PM - 19 January, 2007
Quote:
if you watch what it does, sometimes it wont put it exactly on what you want. So say you have it set at 92, some songs will read in at 92.3 or 91.8, so those are the ones that would have the problem.
I def have noticed this on mine. I set mine at 95 and its as uniformed as I have seen them. 92 to me is too low. Leaves to much room for error.


Read the documentation. It explains why the range is +-1.5dB.