Does anyone do separate mixes just for MP3/Ipods/etc?

Hi Billy I mixed an album a few years ago for the teenage market in Indonesia. They were listening to the music mainly on IPods and earbuds etc. I found that separate mixes were not really necessary. I was able to cope with the slightly different requirements purely in mastering.

They loved the mixes but said the bass end was in fact a little strong for the earbuds and all I had to do was just tailor the bottom end a little differently in mastering and also smooth out the top end as well. I tested the mixes by transferring the music to my son's IPod and checking them on the actual device that would be used.

This project did make me realise though that you do in fact have to check your mixes on headphones and earbuds if there is any chance that the material is going to be heard that way. I never really used to ever worry about how things sounded on phones or earbuds but after this I take them into consideration more.

A great mix will transfer well into any environment so I dont believe doing separete mixes is really necessary. The only time I do mix differently is if I am doing a soundtrack for a doco for TV and I know that in many situations the audio is going to heard on fairly small TV speakers most of the time. Then I find that some parts simply get lost and others can be over emphasised a bit so sometimes I make mix changes to make that work better. But I always do an album mix for my own benefit or for any demos etc.

I say we let the guys with iPods and MP3 players figure it out when they rip it, or with the on-board EQ in the MP3 player itself.

I don't try to "dumb down" my mixes so they sound good on crappy ear buds.

skullsession


I say we let the guys with iPods and MP3 players figure it out when they rip it, or with the on-board EQ in the MP3 player itself.

I don't try to "dumb down" my mixes so they sound good on crappy ear buds.

 
Well, tell that to the hit makers lol.      The truth is, if one listen's reallllly carefully to a lot of modern hit music, the sound is awful to begin with.  It's just that a great mastering engineer saved things in the end.
 
I've been really analyzing/studying a bunch of big hits/styles lately with the sole purpose of trying to figure out what frequencies and degree of compression is being used, and frankly it's a lot of smoke and mirrors.  There's so much focus on the lower mids to mids area, so of course that's easy to "fit" in those earbuds.
 
I agree one shouldn't 'dumb down' a mix, though.
 
 

ba_midi

We all know that the earbuds/headphones being used on ipods and the like tend to lose the bottom end and seem to have an over emphasis on the high end.
 

I've read here (in more than one post, I'm pretty sure) that mp3 compression is most easily distorted by the higher frequency content.  While I have not ever experimented, it would be interesting to try the following.  Convert a "mastered" track to mp3, using a typical compression that gets you roughly a ten-to-one reduction in file size ... in other words, we want a typical amount of "lossy-ness".  Then, put that same track through a low-pass filter with cutoff at, say, 15K.  Reconvert to mp3 and see if the filtering offers an improved mp3 result with fewer artifacts.  Try higher and lower cutoff settings as well.

dmbaer

ba_midi

We all know that the earbuds/headphones being used on ipods and the like tend to lose the bottom end and seem to have an over emphasis on the high end.
 

I've read here (in more than one post, I'm pretty sure) that mp3 compression is most easily distorted by the higher frequency content.  While I have not ever experimented, it would be interesting to try the following.  Convert a "mastered" track to mp3, using a typical compression that gets you roughly a ten-to-one reduction in file size ... in other words, we want a typical amount of "lossy-ness".  Then, put that same track through a low-pass filter with cutoff at, say, 15K.  Reconvert to mp3 and see if the filtering offers an improved mp3 result with fewer artifacts.  Try higher and lower cutoff settings as well.

Actually - you don't even have to go that far.   Just load a mastered WAV file into an audio editor (like Audition 3.x for example) and notice the waveform and FFT spread.   Usually we'll see a normal slope down but with full range (ie, up to 20KHz and higher).
 
Take that same track, only it's MP3 version, and load it in.   You'll immediately see a steep cutoff at 16KHz.  Some even lower (depending on the source).
 
One thing I definitely notice in many 'hit' records/mixes is that the high end is reduced dramatically- and probably intentionally - due to the "mp3" world out there.
 
It's easy to hear the range from about 7K up to 12K is usually attenuated heavily - and then another attentuation in the 12k-16k (or higher) area.   Almost every record I've been checking has these spectrum attentuations.
 
I even find I am mixing a bit differently lately to get 'in the pocket' for MP3 'sound', though not drastically.
But there's no doubt that those records/mixes with those freqs attenuated sound better on mp3 players (earbuds) and still have enough top-end for monitoring on normal speakers.
 
Some of this, of course, is the way the ear hears freqs anyway - and from about 7K up tends to be harsher sounds in modern/pop/hip-hop/dance music. 
 
Another example of "mixing for the delivery system" is having so called "club" mixes.   One would think they would pump the bass more to get heavy sub-woofer response.  In fact, a lot of club mixes have attenuated sub frequencies knowing that the club/DJ will add lots of low end / sub EQ'ing anyway.
 
 

Convert a "mastered" track to mp3, using a typical compression that gets you roughly a ten-to-one reduction in file size ... in other words, we want a typical amount of "lossy-ness".

 
I forgot to say - I actually have done some of that experimenting.
 
I have (legally) the instrumental track and separate vocals only track - in WAV format, of Lady Gaga's "Just Dance".
It's a great track, and actually very simple (when you listen to just the instrumental).   Very solid - but obviously also mixed with 'some' thought about the varoius formats it will be delivered to, including MP3.
 
I've taken that track (instrumental) and analyzed it, converted it to MP3, experimented with 'radio-like compression' and more.
 
The interesting result is that the frequency range is actually, as I've been pointing out, somewhat limited.   There are really no blatant "high end" signals (ie, most of the sounds above around 8K are attenuated).
 
The spectrum might show full range up to about 16K on the MP3, but the down-slope begins around 3-3.5K.
 
 

skullsession


For the sake of posterity...please don't unless it's an alternate mix, specifically for a particular medium - and the REAL mix actually sounds good on good speakers.

I've got to believe there will come a time again when people will actually listen to music on good speakers in their homes.

Well, for as long as I can remember (and I used to be in the music biz for a long time), it was always the case to do different mixes -as I've mentioned:  Radio Mix, Club Mix, TV mix, etc.   So it's not unusual in any event.
 
I agree that "THE" mix should be as good as one can get it - and not purposely mixed for any one delivery system.  Just a good old mix, period.  BUT, that's not really the end of the story in the 'real world'.   Delivery systems matter.  And they change (though now we'll be in the Internet delivery mode for quite some time, I'm sure).
 
I'm not suggesting one "dumbs down" a mix (to use your term).   But to consider the delivery system, and perhaps provide a focused mixed for mp3 delivery, as one example, may not be a bad thing.
 
After all, it really is a matter of finding that balance between "full bandwidth/spectrum" and "just the right spectrum" that fits well.  
 
IOW, I'm not suggesting one mixes solely with the MP3 world in mind, but to consider some balance so that the product is realized in all delivery systems.
 
 

I've got to believe there will come a time again when people will actually listen to music on good speakers in their homes. 

 

Hmm, I'm not convinced of that

ba_midi

Convert a "mastered" track to mp3, using a typical compression that gets you roughly a ten-to-one reduction in file size ... in other words, we want a typical amount of "lossy-ness".

 
I forgot to say - I actually have done some of that experimenting.
 
I have (legally) the instrumental track and separate vocals only track - in WAV format, of Lady Gaga's "Just Dance".
It's a great track, and actually very simple (when you listen to just the instrumental).   Very solid - but obviously also mixed with 'some' thought about the varoius formats it will be delivered to, including MP3.
 
I've taken that track (instrumental) and analyzed it, converted it to MP3, experimented with 'radio-like compression' and more.
 
The interesting result is that the frequency range is actually, as I've been pointing out, somewhat limited.   There are really no blatant "high end" signals (ie, most of the sounds above around 8K are attenuated).
 
The spectrum might show full range up to about 16K on the MP3, but the down-slope begins around 3-3.5K.
 
 

One other thing occurred to me last night.  I've also seen advice here (from Bitflipper, if memory serves) that tracks destined for mp3 conversion should be maybe 2db lower than the max level you'd use for a CD track.  I believe the reason was that mp3 distortion could effectively cause high signals to clip.

ba_midi

One other thing occurred to me last night. I've also seen advice here (from Bitflipper, if memory serves) that tracks destined for mp3 conversion should be maybe 2db lower than the max level you'd use for a CD track. I believe the reason was that mp3 distortion could effectively cause high signals to clip.

 
Heh, well that is good advice, too bad most of the recording industry isn't listening. 
 
I belong to some of these sites that promote/sell hit tracks in both wav and mp3 formats (legally mind you).   You should see the wall-to-wall waveforms even on the MP3s.   The only diff between the "to the max loudness" of the wav vs the mp3 is the normal slope/cutoff about 16k.
 
 

I probably should have used the term "peak level" rather than "max level".  This wasn't a comment about overcompression.  So, if your CD-bound track peaked at -0.5db, then your mp3-bound track maybe should peak at -2.0 or -2.5db.  But then you probably understood what I was trying to say anyway.