OK, so assuming you start with an MP3 on your computer that you wish to remix in a DAW that converts it to .wav, do you add the Limiter on the wav and convert, or convert it to mp3 and then add the Limiter.

I've tried all the options, but I would like to know how other people do it.

Many thanks!

 

yes.

This would be incredibly daft cuz you're going

mp3 > WAV > edited mp3 > limited mp3 = triple (re) compression.

Or wynaut lower by 3-6dB and keep it there?

 

I dont know why I do this. I try to make all the imported audio in my project 48/24 so the DAW isnt doing any on the fly converting while Im working. I use db poweramp music converter and convert any external audio (like acapellas and vox) to the same sample rate and bit depth as the project. Or I just consolidate/freeze it once I get it in (like 16bit samples). It just seemed like common sense to me. But I never asked or looked it up. Maybe Im doing it wrong? Im just trying to squeeze every last bit of CPU out of my setup as possible. And my workflow just developed like that.

 

That is exactly the reason I asked, I found a work flow that I think works, but I'm never 100% that I'm doing it right!

I know people usually respond to something like that , with if it works carry on doing that, but it's the reason I'm never sure.

Thank you for explaining your workflow Sub

 

Ok thank you.

 

We are not talking about the loss when you converted the mp3 to WAV, right? Cause when it happened you sure lost some of the original source material (before the mp3 compression), if it was a flac file for example, you wouldn't lose nothing when converted to WAV. it´s like you getting a cheap JPG importing it into Photoshop to compose a higher bitmap (say 300dpi) while the jpg is at most 100dpi, something would not fit right this way, you get me?

 

No. I understand that. Here is an example of what I mean.

Let's say I have 100 mp3 tracks on my hard drive (some could be 80's tracks that were very poorly mastered originally) I wish to make a megamix out of, adding drum loops and possibly a drum machine (Which I process on separate channels) My DAW converts those tracks to wav files. I then progress with the mix and save it. This leaves me with a .wav file of the finished mix which could be as much as 3gb. I need to convert the finished track back to mp3.
So really my question is, would you add a limiter to the .wav and then do some guess work about how much volume to remove off of the mp3 or convert it to mp3 and then add the limiter.

 

If you are worried about ISP during conversion then leave more headroom in the *.wav file before conversion. There are also plugin that let you audition the conversion and let you see the peaks generated in real time (a lossy compression preview with metering) You might be surprised how much headroom is needed if you want to avoid ISP on some material, and it really is target bitrate dependent. If your target bitrarte is below 196Kbps though - there isn't really a point and it it is 320Kbps you might as well generate a FLAC the size difference will be very small.

 

I need to look to see if the device I wish to play the mix on can read FLAC, but if so, there is no loss on FLAC is there?

it does indeed! :d


Wav = 802 mb
Flac = 565 mb
MP3 (320) = 182 mb

 

Last edited: May 12, 2017

No loss of any kind with FLAC. And it even supports 24-bit. WavPack supports 32-bit even. Just a tad bigger file. I use WP for archiving.

I would do it this way: convert the MP3 to 44.1/32bit with LAME - command line works best, but I guess Foobar2000 could do it, too, as it uses LAME. You could upsample the files to 88.2/32bit for mastering as mastering sounds cleaner at bigger frequencies. Upsampling and downsampling with Voxengo R8Brain free works flawlessly [and it's easy to do]. Use some kind of a metering plugin to equalise the loudness of these files. Best to use R128 capable [LU, LUFS] metering for that, or even better - ears. ;) If you feel "ard ko" you could also equalise the frequencies of every file a little with a graphic EQ like Volko QG, so they will sound more uniform. Graphic EQs tend to work faster with this type of things. Then use a limiter on the master to shave off about 3-6dB [threshold ~ -3 to -6dB], but set the ceiling to about -0.5dB at least if you plan on converting to MP3 later. Render the mix to 88.2/32bit WAV, no dithering. Downconvert with R8Brain [it auto-dithers to 24 or 16 bit audio with gaussian dithering] to 44.1kHz/24bit and encode with LAME to MP3. Well done! You can light up a spliff now and listen to your mix compilation. Enjoy!

Cheers!

P.S. If you don't feel like equalising loudness of the files individually, you can use this little helper to equalise them in 5 minutes: http://r128gain.sourceforge.net/ Extraordinary little program!

 

Last edited: May 12, 2017

Using a regular audio editor/daw you can't add limiting to a mp3 file. Even if you drag & drop an mp3 file to a daw, the daw automatically converts your file to a loseless format, all the processing is done to that format then when you export again to mp3 the file is re-encoded. Think of an mp3 like brand new puzzle pieces in a small box - the daw needs to "solve the puzzle" convert all pieces together in a large file before processing them. So regardless if you see temporary files or not, rest assured your audio processor always works on a uncompressed format.

Things to take in:
- there's no need to separately convert your mp3's to wav before adding them to daw, as the daw will always do it for you automatically
- no audio editor/daw will process an mp3 file (except trimming segments out) without reencoding. every time you edit an mp3, when exporting the file you basically re-encode a temporary wav made by your daw from your original mp3.
- it is not possible to dynamically process an mp3 file (aka add a limiter) unless you play that mp3 file with a media player that supports plug-ins, use a limiter plug-in on that media player and internally record your own soundcard while that mp3 file is being played. But even in this situation, you apply the processing to a wav decoded on the fly and reconvert to mp3 when capturing so whatever you do, you end up processing straight uncompressed data and recompressing back to mp3.

 

Last edited by a moderator: May 12, 2017

So thats what those weird files in my ableton cache folder are. Thanks for enlightening me

 

No matter how much I think I know, there is always so much more to learn!!!

I don't confess to knowing everything you've said, but I'm going to have lots of fun dissecting and learning it.

Thank you, some valuable info there.


In fact, i want to say thanks to all who have helped. I have a lot to read up on and learn now, and add it together with what i already know. Hopefully It will leave me with a definitive way of doing things.

 

Ozone 7 advanced has a codec preview

 

Oh, and if you're converting MP3 > WAV , you should find a converter that decodes in floating point , then reduces gain, otherwise you're clipping already (at the conversion stage) before you even start remixing (since it's probable that your MP3s clip). Saving to 16b (after gain reduce) is fine.

 

We're talking about MP3 files here. They might have a cutoff of 20k, or worse, so it's futile to 'master at 88k'. This is also futile in general. (but 32b FP is ok/good).

If you're doing mastering at 88k AND sounds 'better', there's something wrong goin on.

ANd if you're doing SRC, you're at the mercy of your SRC algo, which can hurt quality, or introduce unwanted artefacts. See also https://audiosex.pro/threads/what-exactly-is-oversampling-and-do-i-need-to-use-it.20157/#post-131883