Mixing/Mastering for MP3's

[Tidwells@aol.com]

I'm looking for advice on creating the best possible sounding MP3's of a some songs I've been mixing & mastering in DP 9. In the past, I found that DP made better sounding MP3's with the L.A.M.E. encoding engine if I dithered the mix down to 16-bit first using a Masterworks Limiter on the master bus. After doing this and then listening to the MP3's in iTunes, I'm noticing that the lead vocals (which are panned dead center) sound weaker and are now overwhelmed by their stereo reverb applied in the mix plus other tracks. So I'm going back to the original mix and boosting lead vocals by 2 db with trim plug-ins while cutting vocal reverb about 15%. Then I re-bounced to MP3 and it sounds "more normal". Has anyone else run into this? It seems like the MP3 conversion process is producing a "stereo widening" effect and de-emphasizing mono information slightly.
For L.A.M.E MP3 bounce settings, I've been using these settings:
Channels: Auto
Stereo Mode: Joint Stereo
Output Sample Rate: Auto
Encoding Mode: Constant Bit Rate
Encode Quality: High
Bit Rate: 256 kbps (High)
And I checked the box to Filter frequencies below 10 hz.

Anyone have any advice? I just want the MP3's to sound as close to the original 24-bit 44.1hz mixes as possible (of course). I have been setting the limiter ceiling at -1 db. I read somewhere that this is a safer level for mixes being converted to MP3.

Doug

[FMiguelez]

After doing this and then listening to the MP3's in iTunes, I'm noticing that the lead vocals (which are panned dead center) sound weaker and are now overwhelmed by their stereo reverb applied in the mix plus other tracks. \

Are you absolutely sure you're actually hearing all those changes in your music you say? Are you monitoring through the same chain or is something else possible? I've never heard of a modern mp3 CODEC that changes the mix like that!

Honestly, with the most up-to-date AAC or MP3 converter, be it Apple's or LAME, at 256 kbps, ALL my mp3 or AAC conversions sound IDENTICAL to the 24 bit 48 KHz master to me, and I could most likely not pass a blind test comparing them under normal listening conditions (in my studio) to save my life. Could you?

My friends can't either when I've tested them (and embarrassed them).

Are you sure you're using proper settings? The ones you mention seem perfectly fine to me. LAME is as good as it gets (however I prefer AAC). There should be NO difference.
Have you made AB blind comparisons?

This should be no issue now a days...

[mikehalloran]

I do floating point, highest quality. It takes a pair of headphones for me to hear any difference and it's minor when I A/B'd. The mix is unchanged. I've no problem with 256 which is the iTunes default I've set for for my iPhone.

I neither bounce to nor convert to mp3 inside Digital Performer. Life is just too short. A one hour mix down with conversion to mp3 can take 30 minutes or more for me. Instead, I mix to .wav and convert the mix in a dedicated converter. I have a few including DSP-Q but my favorite is TwistedWave—it can convert a 2 hour .wav mix to mp3 floating point highest quality an 2–4 minutes or less.
https://twistedwave.com/mac

[HCMarkus]

This should be no issue now a days...

+1

[monkey man]

Stereo Mode: Joint Stereo

There's your problem right there, Doug.

I never use "joint stereo"; it messes with imaging and might explain the anomalies you experienced with that / those centre track/s.

It's a space-saving feature who's gains are minimal. If you're going 256 or 320k it's like shooting yourself in the foot.

This is all just MHO. YMMV.

[mikehalloran]

Stereo Mode: Joint Stereo

There's your problem right there, Doug.

I never use "joint stereo"; it messes with imaging and might explain the anomalies you experienced with that / those centre track/s.

It's a space-saving feature who's gains are minimal. If you're going 256 or 320k it's like shooting yourself in the foot.

This is all just MHO. YMMV.

I believe that our simian friend has nailed it.

[HCMarkus]

Monkey earns his daily banana, film at 11:00.

[sayatnova]

This is good info to know. Just for educational purposes to others who will want to learn from this thread: if not «joint stereo», which is the best option to choose to keep quality high and not mess with the stereo field?

Thank you!

[FMiguelez]

I thought the stereo setting in the MP3 encoder, at least from DP, means the same thing as having 2 mono files (deinterleaved) or one stereo file (interleaved). It doesn't have anything to do with stereo levels or anything better or worse than the other, it just is a different "presentation" of the same file... Like having regular stereo VS MS. Or am I completely wrong here?

If I have time I'll test this. It should be easily doable and informative

[mikehalloran]

I thought the stereo setting in the MP3 encoder, at least from DP, means the same thing as having 2 mono files (deinterleaved) or one stereo file (interleaved). It doesn't have anything to do with stereo levels or anything better or worse than the other, it just is a different "presentation" of the same file... Like having regular stereo VS MS. Or am I completely wrong here?

If I have time I'll test this. It should be easily doable and informative

That's it basically.

Joint stereo is a scheme that matrixes the information common to both channels. This may reduce file size in 128 or lower. Despite best intentions, it's notorious for changing the sound. In the past, when storage space was at a premium, it was considered advantageous to do this. That was a long time ago.

Although often lumped together with M/S when described, that's a false equivalency. M/S begins as two mono signals. By matrixing the two, you wind up with true stereo. One side is M+S while the other is M-S. (which is L and which is R depends on the initial polarity of M).

The only thing common to the two is that a matrix is involved. You don't even need a computer to set up M/S. A box, a few connectors, some wire and a potentiometer (optional) is all that's needed. The pot lets you you change the spread by varying the amount of S in the matrix. In fact, you could get away without the box but that would be ugly.

[funkyfreddy]

I usually do my mp3 conversions in iTunes. Am I missing something?

[mikehalloran]

I usually do my mp3 conversions in iTunes. Am I missing something?

Not really. Other converters may give you more options but if you don't need them, that's ok.

[EMRR]

I usually do my mp3 conversions in iTunes. Am I missing something?

After a bit of playing with LAME, I could never get a result that sounded as good as just using Itunes.

The MFIT spec is worth reading. Probably the most important point is to never have peaks over -1dBFS, to avoid mp3 conversion distortions. Obviously that's an average recommendation, wouldn't apply to everything, but how would you know? Plus, given LUFS standards in streaming and broadcasting, anything that hot gets turned down anyway. DP Masterworks Limiter requires a setting of -1.1 to land at -1.

The mastering house I've used the most has always recommended keeping bit depth as high as possible, right up to mp3 conversion.

Check this if you haven't seen it:

https://www.loudnesspenalty.com

[funkyfreddy]

Thank you Mike and Doug!

My ears are usually shot by the time I get to the mp3 stage from listening to the same piece of music over and over while editing and mixing LOL... and I don't usually sell mp3's. They are just quick and dirty necessities that I e-mail to clients before I send them the full bandwidth files.

I tried a few times with an earlier version of DP adding LAME and it always gave me issues. I haven't found any problems with iTunes so far as an mp3 converter so I will stay with it.

[mikehalloran]

DP uses LAME but it installs automatically since DP 8.03—same with DSP=Q and TwistedWave. Which version(s)? I neither know nor care.

Apple uses their own mp3 encoder.

My only issue is that, in DP 9 & 10, the conversion seems to take forever.