Mastering for MP3 - any tips?

  • Thread starter Thread starter zoetrope
  • Start date Start date
Z

zoetrope

New member
I will be mastering some wav files for a compilation CD to be released on MP3.com
:eek:
I know.
Anyway, I'm looking for tips to maximize sound quality given that I'm ultimately converting the wav files to MP3 files at 128 kbs. Are there specific things people do to minimize the damage on the trip from "CD quality" to "almost CD quality", also known as "frying pan" and "fire" respectively.

Any advice greatly appreciated.
 
I haven't tried this yet, so it may work spiffy or not worth a crap - but eventually I WILL try it, so if you want to do it first (yeah, like nobody's thought of this yet) be my guest...

Burn your MP3's at the bit rate you wanna end up at, no EQ or anything, just good-sounding wav files, work well on your system, etc - then, take the MP3's, run them thru the same system, preferably in a way that you can quickly A-B the MP3 with the wav version - tweak the EQ on the MP3 track til it sounds as close to the wav file as you can get it -

assuming you're using a DAW for this, such as sound forge or Samp, etc - save the EQ setting that worked best, as a preset - call it MP3Burn1, or something you can remember.

Now, go back to the wav files, run them all thru this EQ preset as wav files, and THEN MP3 them and see what it sounds like - with any luck it should be fairly good, if not great (yeah, I know, they're still MP3's...)

Tweak the EQ preset as necessary, untill you can just run a good sounding wav file thru the preset, convert it to MP3, and like it.

Seems like this oughta work, bur remember I've not tried it yet.

Then again, if you're sending these mp3's to Amazon or one of the other indie helpers, maybe they ALREADY have a tweak that makes the MP3's sound (almost) good enough to go on a CD... Steve
 
my best advice is to make it sound as good as possible BEFORE conversion, and to use the highest-quality encoder you can (many of the higher-quality encoders take LONGER to encode, if that's any help)-i tend to use the 2nd-generation fraunhoffer codec, and it's done me right. not much trouble with quality on the high-end, as many mp3 encoders have, but a little difference on the bass (note: "difference" does not mean "sounds bad"-if i were to listen to my 192-kbit recordings of a godflesh album i ripped without A/Bing it with the original, i would say it sounds great-but i can tell the difference in comparison. i've done my own recordings at 160 with this codec, and i have a damned hard time telling cd's burned from the mp3's from the original. that said, it's just me and a guitar on those recordings, no thick bass.
 
pink noise test

knightfly
I recorded (generated) a few seconds of pink noise, did a frequency analysis, converted this to MP3, did a comparison frequency analysis, and the results are nearly identical. There is a sharp drop off in the high frequencies, but this is a low pass filter that can be set when creating the mp3. Otherwise, the two were essentially the same.
So there you have it.
 
Not quite - If I'm not mistaken, isn't the MP3 thing about "throwing away stuff that would be masked anyway" ? - If so, I doubt that a steady state pink noise would be encoded the same way dynamic music material would be.

If you're still feeling curious, You might try the same experiment with 3 distinctly different genres of music, compare the resultant FFT's and see if the high roll-off is STILL the only diff... Steve
 
Wow that really sucks. I hate mp3s. I don't even download (steal) music off the internet anymore because they sound like such shit. You had to A/B it to tell the difference? Why can't you use a higher bit rate than 128? Thats pretty low.
 
mp3.com

The CD will be on MP3.com, which limits bitrate to 128 (unless you sign up for their premium service etc.etc.).

The problem with my experiment, as pointed out, is that pink noise doesn't leave anything for the MP3 to mask. But for every song the EQ spectrum that results from the conversion is going to differ from the original spectrum in a different way. While narrowing it down to genres will limit the differences somewhat, I still think you would have to create a new EQ set up for each song just to undo what the MP3 conversion has done, and even then you will only have limited success since the conversion will now mask the newly EQed file in an unpredictable way. For any given Wav, each time you correct the EQ the conversion will turn it into something new to correct. I suspect that each time you do this you will be no closer to the original spectrum, regardless of how many EQ fixing generations you have done. Admittedly, that is a guess, I haven't tried it. I will and see what happens.

Anyway, any other thoughts on the subject (other than MP3 sucks dude :eek: ) would be appreciated.
 
I don't think there's really anything else you can do except start with the highest quality possible and use the best encoder as possible.
 
Hey Tyler - long time no hear, where ya been?

Looks like you still need to read as carefully as I write (helps understanding, at least I HOPE it does)

What I said was, "quickly A-B the MP3 with the wav version - tweak the EQ on the MP3 track til it sounds as close to the wav file as you can get it - " -

Note the term, "as close to" - Nobody said the MP3 was going to be "as good as", the point of this would be to make the MP3 into as CLOSE a copy of the WAV file, EQ wise, to not need further tweaking in order to be as good as the bit rate would allow.

Just a clarification, hope you had a good Christmas... Steve
 
knightfly said:
Hey Tyler - long time no hear, where ya been?
Click to expand...

I been around, reading mostly. I have no idea why I'm back, but I am.

What I said was, "quickly A-B the MP3 with the wav version - tweak the EQ on the MP3 track til it sounds as close to the wav file as you can get it - " -
Click to expand...


I understand what you're saying but wouldn't that mean you'd have to go through another generation of wav and mp3 in order to do this, and even if you didn't have to do that, like if you had an mp3 editor, which I don't believe exist, or aren't very good, wouldn't doing it just raise the noise floor, or the raise frequencies just be removed again?

I really don't know much about mp3s so I'm just guessing here.

Note the term, "as close to" - Nobody said the MP3 was going to be "as good as", the point of this would be to make the MP3 into as CLOSE a copy of the WAV file, EQ wise, to not need further tweaking in order to be as good as the bit rate would allow.
Click to expand...


I understand.

Just a clarification, hope you had a good Christmas... Steve
Click to expand...

You too! :) And New Years now.

-Tyler
 
you're going about this all wrong.
you can't EQ out the artifacts of MP3 encoding. they are ARTIFACTS - i.e. limiting of stereo seperation, phase distortion, brief transient cuts, etc. an MP3 encoder does not say "okay, cut 1500Hz by 2db and that will save us some space." it just doesn't work like that.
the only thing you can do is get the source material sounding great, then use the best encoder possible (currently the LAME free encoder, if you read enough of the user guide to use all the command-line switches correctly.)
 
Bleyrad, that was (to me, anyway) kind of a given - I wasn't trying to treat the artifacts, other than the one that makes the apparent EQ sound different. My only aim in this discussion was to make it so you didn't feel the need to grab the EQ when switching from MP3 playback to CD playback, or vice versa -

Tyler, read this line again - "go back to the wav files, run them all thru this EQ preset as wav files, and THEN MP3 them and see what it sounds like - with any luck it should be fairly good, if not great (yeah, I know, they're still MP3's...) " -

If you follow the logic, I was meaning that you should find out what EQ works best to keep the MP3 version sounding similar in EQ to the original wav version, and THEN - THEN - make those adjustments to the ORIGINAL wav file BEFORE it gets encoded to MP3 the FIRST time - follow?

And, Bleyrad's right, you still can't make a silk purse out of a sow's ear, but you might get out of having to tweak the EQ just to listen to the pig squeal... Steve
 
of pigs and purses

you still can't make a silk purse out of a sow's ear, but you might get out of having to tweak the EQ just to listen to the pig squeal
Click to expand...
Hey, how did you know I was recording an album of singing pigs. I can't keep anything a secret these days.:mad:
 
knightfly said:
Tyler, read this line again - "go back to the wav files, run them all thru this EQ preset as wav files, and THEN MP3 them and see what it sounds like - with any luck it should be fairly good, if not great (yeah, I know, they're still MP3's...) " -

If you follow the logic, I was meaning that you should find out what EQ works best to keep the MP3 version sounding similar in EQ to the original wav version, and THEN - THEN - make those adjustments to the ORIGINAL wav file BEFORE it gets encoded to MP3 the FIRST time - follow?
Click to expand...

Okay, sorry, Looks like i need to read things over 3 times now. I follow now.
 
You must log in or register to reply here.
Back
Top