Headroom for mp3 conversion or what?

Crescendo

New member
Do I master at a too high level (-.1 DBFS) or why do the dynamics of my material get f*cked up on some occasions when uploaded to internet? Is leaving headroom in the mastering stage something to consider? I uploaded something myself first, okay, sounds fine at this stage, a guy then downloads it, then re-uploads it, and now there's tons of swaying/compression all over the place. It was never there before that. I mean taking a few things into consideration I find hard to believe that simply re-uploading an mp3 would affect the sound this much, but it's possible isn't it? Or is it just due to bad mixing?
 
Don't upload MP3s! Where are you putting them? Some sites (like soundcloud) apply their own compression algorithm which can (depends on the file and how much compression is on it already) really screw up the sound. D/l it, then reupload it and you're applying yet another layer of compression.
 
Yeah it might've been one of those alorithms but it's weird how it sounded okay when I played it on the site, which was clyp.it by the way.
 
Anything you upload anywhere, try to upload it as a wav file.....if at all possible.
Im saying if possible because some places just wont let you. For example here on HR, you can't upload wav files.

All the music sharing sites do their own compression so if you upload an already compressed file (mp3), they'll do their 'magic' on it and completely ruin it.
:D
 
It's good to leave a little headroom regardless of who does the file compression.

Soundcloud takes whatever you give it and adds their own compression, so it's best to upload wav files. It's also good to upload uncompressed files to Bandcamp, which compresses them for streaming and allows downloads of various quality according to the listener's preference. ReverbNation only takes mp3 files with a size limit, so you have to compress them yourself, but you can use whatever amount of compression keeps the file under the limit.
 
-0.1dBFS is -- I mean, if you're trying to avoid intersample peaks on anything, much less weird MP3 artifacts, yeah, that might be a bit hot. Especially if it's slamming that number with regularity. At least a half dB for just about anything that's hitting hard (which still, in the grand scheme, isn't anything) and maybe .75-1dB down if you're exporting directly to (or know you're converting to) MP3. Granted, we know we're not looking at getting in the good graces of the audiophiles with MP3 anyway, but it's nice to not try to kill people's (usually non-stellar) converters.
 
When I know I'm going to ReverbNation with a tune, I purposefully set the limiter's output to -.5 and turn on the intersample controls. Normally, I still run at -.3, but I'm no pro at this stuff. I just know I don't get the clipping if I turn it down a touch for Freemake's MP3 conversion from .WAV.
 
So I remixed and remastered a year old EP (my bass tone back then wasn't that great and crunchy but I left it like that anyway) and I mastered at -0.4DBFS.
https://soundcloud.com/kres90/cres-vf-vega-ep
I uploaded it as wav. Do you guys experience any of these dynamics problems I talked about earlier, when you're playing this on your system?
 
To 0dBFS? No. My limiter is set at -0.4 as stated earlier
I test-listened to music on this one guy's laptop and the integrated soundcard's settings were so stupid, tons of compression. It's like engineers ignore the fact that people listen to music a lot on computers these days, and don't just use VoIP. But the effects were more prominent on metal songs. I hate the fact that distorted guitars take so much space. But sometimes I feel like mine take more than they should, hence the original question about getting tons of swaying in some circumstances. Still, it's not like my guitars peak at 0dBFS. That would be too loud obviously.

Yesterday I read that Soundcloud recommends leaving 3dB headroom. I should upload something as a private track, normalized to -3dBFS, and do more investigation.
 
I limit/normalize at the final step to -.1db , and release a lot of music online via MP3 and the original WAVs. No fidelity issues. Are you sure your process is correct and that the final file is actually peaking where you tell it? For example, I open the final WAV in Soundforge and run a quick analyzer on it to verify the peak and RMS values to be sure they're close to what I'm seeing in the DAW's meters. They are not always an exact match, the fault of the DAW (Sonar). But knowing this and checking for it eliminates any baked mixes getting through.

Fwiw - while doing track level work and/or at the first couple mastering stages, I obviously leave more headroom than that.
 
What are your mastering stages if you don't mind me asking?

Yes, of course. The peaks are where they're supposed to be (RMS values are a bit lower than in DAW). I also have SoundForge and I always open my masters on SF, regardless of whether I'm going to do fade-outs/fade-ins etc.
I think part of the problem was that I had a brickwall HP filter on my kickdrum, which caused resonance, which in turn caused a really long low-end sustain that I couldn't hear at any stage, but it was definitely eating up the headroom. Also, I did not HP my bass guitar. I hope that my material translates to other systems better now.
 
There's some variation in my process, and I recently had a thread in this section addressing (and finalizing) my mastering chain. I do some soft compression (C6 multiband), EQ, light tube compression (still tinkering with my Puigchild plugin), and some hard prioritized limiting (Waves L3) at the very last step. That comes out of my DAW at -0.2 to -2.0db and in Soundforge I normalize to -0.01 as well as verify the RMS is ballpark. Frankly, my mixes are pretty much the final product and I do very little shaping in the mastering stage on most tracks.

So it sounds like you're not hearing issues, which means your monitoring situation needs some improvement(?).
 
I wish I knew. My multi track recorder seems to blow the volume up over 200% after I master and normalize the song file.
 
Loudness

I'm amazed that everyone still trys to push the limits of mastering in a day & age when most streaming services and radio broadcasters level their output. One guy posted that his limiter was set to -0.01 and he couldn't understand why it was clipping. The loudness war is coming to an end but so many are still on the battle field. My advice is to google or look on youtube for Ian Shepherd. There is an abundance of videos on the subjects of lossy conversions, truncation, intersample peaks and clipping. Every streaming service has its own set of specs and a criteria for uploading to their site. Ozone7 has a great process that eliminates guesswork and lets you preview & meter your file destinations. Otherwise... Melda has a free loudness (not RMS) meter and there are other reasonable offerings for less than $30. Every song is different and there are so many applications that asking for some formula that applies to everything is a bit naive. Research will be your best friend and in today's world, information is free.
 
Back
Top