Downmixing 5.1 to stereo reduces dynamic range?
Mar 9, 2021 at 9:29 AM Thread Starter Post #1 of 20

theaudiologist1

100+ Head-Fier
Joined
Sep 23, 2019
Posts
232
Likes
32
Location
Uranus
So I downmixed one of my albums that were 5.1 to stereo. The original 5.1 version had a dynamic range of 11-13 using TT Dynamic Range meter. However, when I downmixed them, the stereo versions had a dynamic range of only 6-8. Is this normal or is my downmixing software (foobar) bad at downmixing?
 
Mar 9, 2021 at 11:57 AM Post #2 of 20
Downmixing is just mixing tracks and changing gain to some usual values(or to your own preference). In general you want left and right to stay the same, and you'll have some of the other channels at -3 to -6dB before the actual mix with left and right channels.
There is no compression involved. It's just that when you mix different tracks, the quiet passage on one track might not be a quiet passage on the other. So the mix of both might not have a quiet or silent passage at all. I'm guessing that's the reason for those values, although I wouldn't imagine an apple to apple comparison between 5.1 and 2.0, even if a tool gave me some numbers for both.
 
Mar 10, 2021 at 12:12 PM Post #3 of 20
Downmixing is just mixing tracks and changing gain to some usual values(or to your own preference). In general you want left and right to stay the same, and you'll have some of the other channels at -3 to -6dB before the actual mix with left and right channels.
There is no compression involved. It's just that when you mix different tracks, the quiet passage on one track might not be a quiet passage on the other. So the mix of both might not have a quiet or silent passage at all. I'm guessing that's the reason for those values, although I wouldn't imagine an apple to apple comparison between 5.1 and 2.0, even if a tool gave me some numbers for both.
Do you know any good software for downmixing or is foobar already good? Also, what does LFE mean?
 
Mar 10, 2021 at 12:41 PM Post #4 of 20
No, I'm not sure bad ones can exist given how basic the operations are.
LFE is the sub channel.
image2130.jpg
 
Mar 10, 2021 at 4:17 PM Post #5 of 20
Maybe the measurement of the dynamics is done channel by channel individually, and front leaning sound reads as a higher dynamic because the rears are quiet?
 
Mar 10, 2021 at 6:41 PM Post #6 of 20
Maybe the measurement of the dynamics is done channel by channel individually, and front leaning sound reads as a higher dynamic because the rears are quiet?
No, I'm not sure bad ones can exist given how basic the operations are.
LFE is the sub channel.
image2130.jpg
So I did my own mixing (I have 0 experience in mixing) of the 5.1 files in Audacity. I reduced the gain of the FL FR BL BR channels by -6dB and the FC and LFE channels by -9dB. I downmixed and panned the FL and BL 100% to the left stereo channel, downmixed the FR and BR 100% to the right stereo channel, and the FC and FLE I panned center and downmixed to both left and right channels. My own mix has a dynamic range of 11-12. To me they sound good but I don't trust my own mixing. Overall the new stereo tracks are quieter than the 5.1 tracks, and sound better than the foobar-generated one. And yes, I DID put in the LFE channel. Was that a bad idea?
 
Mar 10, 2021 at 7:20 PM Post #7 of 20
A fold down should include all the tracks. Try normalizing up your mix to 95% and see if it will go back to regular volume.
 
Mar 10, 2021 at 8:01 PM Post #8 of 20
A fold down should include all the tracks. Try normalizing up your mix to 95% and see if it will go back to regular volume.
I don't know what a fold down is. I already exported all the tracks. The volume is fine though, but I don't know hot to normalize in Audacity. Do you mean reduce the gain to something like -3dB? That will reduce the dynamic range, though.
 
Mar 11, 2021 at 3:34 AM Post #9 of 20
Fold down is the same as down mixing. Normalizing is raising the overall volume level to just below peak level. Audacity must have a normalizing function. That is pretty basic.
 
Mar 11, 2021 at 4:01 PM Post #10 of 20
Fold down is the same as down mixing. Normalizing is raising the overall volume level to just below peak level. Audacity must have a normalizing function. That is pretty basic.
The thing is I have all my songs as different tracks and don't know how to put them in 1 project. And I think "normalizing" the levels qould still give a lower dynamic range. The first time without reducing the gain I got a dynamic range level of 6-7 just like the foobar one. The second time I reduced the gain of the FL FR BL BR channels by -3dB and I got 8-9. The 3rd time I gave -6dB to FL FR BL BR and -9dB to FC and LFE and got 11-12.

Overall was my mixing good, or is it an artificial way of boosting the dynamic range?
 
Mar 11, 2021 at 4:30 PM Post #11 of 20
Personally, I don’t trust the way apps measure dynamic range.
 
Mar 11, 2021 at 4:52 PM Post #12 of 20
Last edited:
Mar 11, 2021 at 5:02 PM Post #13 of 20
I think the problem might be that the same file might measure dynamics differently at different volume levels.
 
Mar 11, 2021 at 5:47 PM Post #14 of 20
I understand, but overall I followed this:
https://forum.videohelp.com/threads/367078-Need-advice-about-mixing-5-1-down-to-stereo

I followed what post #8 said for my mixing as it explained clipping if you don't reduce the gain. Post #29 also explained clipping a bit. But this is from a video mixing forum. I don't know how it goes in pure audio.
Of course you have to mind the peak level when mixing tracks together. If at a given time you have a full scale signal on one track, and you mix it with another full scale signal at that same instant t, as waves you should end up with the sum of both amplitudes at that point in time. Twice FS, that's +6dB on a system that stops at 0dB. Obviously that would cause digital clipping somewhere if left alone.
But I would imagine that someone going through the effort of coding an automated real time mixer, probably knows that much.
 
Mar 11, 2021 at 6:27 PM Post #15 of 20
Of course you have to mind the peak level when mixing tracks together. If at a given time you have a full scale signal on one track, and you mix it with another full scale signal at that same instant t, as waves you should end up with the sum of both amplitudes at that point in time. Twice FS, that's +6dB on a system that stops at 0dB. Obviously that would cause digital clipping somewhere if left alone.
But I would imagine that someone going through the effort of coding an automated real time mixer, probably knows that much.
So did I do the right thing my reducing the FL FR BL BR streams by -6dB? Also, was reducing the FC and LFE streams by -9dB good or too much?
 

Users who are viewing this thread

Back
Top