Anyone got any expereince or info if there is a difference in how these two calculate the average level of music and which is the most accurate to how we actually experience loudness?

Mazter

Just to close this if it is of interest to anyone.

I just did a small test on some 20ish songs.

first i ran then through mp3trim pro with target -6db

then i listened to them and noticed the differences between softer and more pumping songs. seems the softer ones tend to sometimes be higher in volume to me.

the i ran them all through mp3gain and voila! exactly those i reacted on was reduced a bit.

conclusion: mp3Gain uses a better more audiable realistic to "judge" the volumes in songs which maybe also explains why it uses more time to do the job!

cheers!

Mazter

Hmm, does MP3trim really do volume normalization at all? I thought that setting it to - 6dB just lowers the volume by just 6 dB?

EDIT: I was wrong, disregard this post.

Actually mp3Trim Pro does a great job at it in addition to other neat functions like cleaning up silent parts and fixing bad headers and more.

I am not sure how these program calculate these levels but I assume it has to do with RMS levels (I am a dedicated SoundForge user).

It also allows you to trim manually and/or fade in/out part should you want it.

My conclusion earlier that mp3Gain is better is unfortunately faulty!
It was based on a test made on a too small material.

Last night I ran 2400ish songs through it (yes it took hours) and this morning i did some major comparing on pieces where I know mp3Trim has problems realizing a good listening level (some soft songs compared to harder ones etc).

Both programs perform quite even and both can misjudge but what I found was that mp3Trim is more reliable and makes fewer vital mistakes so I am sorry but mp3Gain just dont match up in the big scheme. Also mp3Trim is faster.

One difference is of course that mp3Gain is free while mp3Trim is commercial but for the amount of songs I am processing (closing up to 70.000 now) i find it worth it.

My personal opinion as usual of course!

I would love to see mp3Gain gain the overhead though because such a beautiful piece of software is what we all need and it should be free

Mazter

As for mp3gain: replaygain


Personally, I would trust replaygain more. I think mp3Trim is using plain RMS and is potentially inferior to replaygain. And that's also the reason for mp3Trim being faster

I doubt mp3Trim uses only RMS since the audible result i compared was overall better to my ears but of course i dont know! Your guess is as good as mine though =P

What i did this morning was to take a number of selected song and run them through mp3Trim.

After that i did an mp3Gain analysis but i did not apply! From these desired changes I could listen to the existing mp3Trimmed material and compare songs/percepted volume to the mp3Gain suggestions and from what I could see I liked the mp3Trim result better. mp3Gain sometimes wanted to do changes which I definately did not agree with.

On most of the material they behaved quite similar though so i was only able to test this on perhaps 6 or 7 songs where they obviously disagreed.

I will do some more tests when i get home from work but i dont expect to see any major difference in the result.

Mazter

Mazter,

how did you used mp3gain? Letting it to modify (amplify) the "global gain" of the mp3 stream at the frame level or just by computing needed adjustements, tagging the mp3 file with those values, and then letting the player to obey to those values at playback time?

You get the first behaviour if you "apply" the album gain, the second one if you just "analyze".

The second option has a VERY MUCH finer resolution compared to the first that operates in 1.5 dB steps.

At least, this is my understanding...


Regards.


Sergio

I must try MP3 trim on some of my music. The majority of my MP3 encodes are Trance music, which usually run like the following:

FULL BELT FULL SIGNAL for 2:00ish
Break (quiet bit) for 0:30 odd
Absolutely full signal again for 2:00ish
Break (quiet bit) for 0:30 odd
Full signal for the last 2:00

Now, say that the breaks are a minute long each, MP3 gain will crank the volume up much more than a song with only very short breaks... Thus, is pretty useless for Trance Singles. In fact, normalizing will give better results.

I'll have to have a sniff at MP3Trim, see how that gets on

Lev,

I think you're doing "track" analysis and adjustements and not "album". Am I wrong??

Sergio

smz,

i am new to mp3Gain so maybe i do somerthing wrong?

i simply add files, analyze (radio) and then apply suggested change.

i use winamp2 for playback

is this wrong?

@Mazter:

It is not *wrong*, but is very likely that it isn't what you're looking for.

Radio analysis will make all the tracks to sound at almost the same "volume", killing the dynamic relationships that exists between different tracks of a typical "album".

That is a very quiet track will be strongly amplified while very loud one is likeley to be attenuated.

This is good for radio broadcasting, where you want a consistent volume along the broadcast (mainly because you don't want people to constantly play with their radio volume knob maybe while driving) but is definitely not an "audiophile" behaviour.

Album analysis, instead, compute a global amplification factor for all the tracks of an album to make the album globally sound at the same level than other albums. And this is what you probably want.

The other thing is that you APPLY that computed gain. This operate at the mp3 frame levles, modifying a parameter, the "global gain" field I was talking about. This parameter allows to adjust the MP3 volume in 1.5 dB steps.

On the other side, if you don't apply, just a tag is written to the files with the computed amplification coefficients written in it as a floating point dB value (much more precise).

The other side of the coin is that while the frame level "applied" MP3 gain will have effect on ANY MP3 player, software or hardware that is, the "tagged" values will have effect only if the player is "replaygain aware".

Personally I use Winamp with the Shibatch mpg123 input plug-in that is "replaygain aware". I'm not sure right now if the standard Winamp input plug-in is replaygain aware or not. Many other replaygain aware players exist, notably the much apreciated foobar2000.
You would make some test with your favourite player...

So, to summarize, 4 options exist with mp3gain:

Album analysis, gain applied
Track analysis, gain applied
Album analysis, gain NOT applied <==== I use this one
Track analysis, gain NOT applied

"track" is the same as "radio", as far as I know...


Best regards

Sergio

Thanks smz

Its all clear to me now...

Yes probably i should have used the album function but there is no way i will rip again 2400 songs for this function so i'll stick with track normalization.

Just wish there was an even more clever way to keep those slow, silent, drumless songs from being boosted too much!

Oh, and this is where mp3Trim seems to do better but i still need more testing.

Thanks again mate!

Good news for you, Mazter!

You don't have to: you can easly undo the "application" of your replaygain within mp3gain an then redo an album analisys.

Cheers!


Sergio

sigh, had i not run them through mp3trim also ....

<sings> you got to live and learn!!! ))

thx sergio!

oops...

You're welcome!


bye!

Sergio

I just followed the advice in your post. I'm using the in_mpg123_118_ot63sse in_mpg123.dll with Winamp 5.02.

I just spent the last four-five hours analysing (not applying) my LAME 3.90.3 -aps encoded MP3s. Result? Nothing that I can tell. I remember noticing a huge difference when I turned on and off ReplayGain with MPC. Here there is no change between enabling ReplayGain and not that I can tell.

I did track gain and set MP3Gain to prevent clipping. In Shibatch's input plug-in, I have "Track gain" enabled. When I uncheck the enable box and hit apply as a file is playing, there is absolutely zero difference in the audio level. I even tried finding one of the loudest songs I had: The Offspring - Gone Away, and played it next to Beethoven's 5th Symphony with track gain enabled. The Offspring song was way louder, so obviously the track gain is not working.

Hell, even changing the Pre-amp value from 0db to -12.0db to 12.0db and applying each time results in no audible difference.

Edit: Nevermind. Winamp must've been using in_mp3.dll. I deleted it and it seems to be working now.