Tinny Audio After Mixing 2 Tracks

[EmpireStrikesBack198]

Hi,

I am having an issue that I am not familiar with and I don't know the name of issue, but I'm sure there is one. It isn't easy to describe the issue, so I attached samples of the problem. I am trying mix Original_1 with Original_2 in Adobe Audition. Original_1 and Original_2 are almost identical, except Original_2 is missing some instruments that Original_1 contains. However, when I mix the two together, the audio quality is degraded and sounds tinny as heard in MixedResult. Also, if you you look the spectral frequency, MixedResult shows a wavy line throughout the audio that seems to have been created after mixing Original_1 and Original_2. I found that I can avoid this issue if I offset one of the Original files, but then they aren't in sync and it is very noticeable. I would like both of them to be in sync.

How can this tinny sound issue be avoided, but still achieve what I want?

Thanks in advance for any suggestions!

[hello_hello]

It sounds like a phasing problem when the two tracks are combined. Offsetting one of them might stop the "phasing" effect. You could fiddle with the offset but maybe never fix it.

If you swap the left and right channels of one of the tracks before combining them I think the channels with the most audio in common will then be in opposite channels (or in other words, separated) so they shouldn't cause any phasing. In fact, it should give you a nice stereo effect. You'll still hear the same thing if you ever play the mixed version through a mono system, but in stereo I think it should sound okay.

[EmpireStrikesBack198]

Thanks for the suggestion and your sample does sound very nice. However, I don't want to ruin the integrity of how the piece was originally mixed. Unfortunately, reversing the channels will affect the original intent. Is there anything else that can be done to avoid the phasing effect?

[smrpix]

Audition should allow you to invert the phase on one of the channels.

[EmpireStrikesBack198]

Audition should allow you to invert the phase on one of the channels.

I'm not sure I know what you mean. Invert one channel on one track, but what about the other track? Wouldn't there still be a phasing issue on the other channel?

[hello_hello]

I haven't looked at the waveform closely, but the idea would be to get them to match up exactly, so it might be possible by delaying one of them by a fraction, and maybe if you can view the waveforms side by side, match them up that way.... and it could only require a few milliseconds here or there, but for various reasons it mightn't be possible to match them exactly (I don't know how the audio was created).
Assuming they're supposed to be "combinable", delaying one track (positive or negative) by a few milliseconds might fix it, and if a very small positive delay makes it worse, try the other direction to see if a very small negative delay makes it better etc. It might just take some trial and error...... Or putting them in separate speakers might be easier.

As an extreme example if you take an audio track and make a copy, then phase reverse the copy before combining them, while separately they'd sound the same, as they're the exact opposite of each other, when you combine them they cancel each other out and theoretically you get silence (in your case I think phase reversing one track will make it a lot worse, but if Audition will do it, it can't hurt to try).
When you have two identical sounds which are slightly out of phase combining them can cause that "phasing" sound. You could probably take any audio track, make a copy, delay the copy by a small amount, then combine the two and you'll hear a similar thing. Well.... I did..... I'm pretty sure this is the effect you're hearing. I used a 10ms delay to make it obvious.

[hello_hello]

One other thought.... if the DC offset for each track isn't the same it could possibly produce the same phasing effect. I can't say I've thought about it until now but......

DC offset isn't hard to understand. It's explained here: http://wiki.audacityteam.org/wiki/DC_offset
If Audition has a DC Offset Correction option, and I'd be surprised if it doesn't (I've never used Audition myself) maybe try it on the individual tracks before combining them. You never know.....

[EmpireStrikesBack198]

Thanks for taking the time to try to figure this out and offer suggestions. I have tried to delay one slightly, but I had to delay it more than I would have liked to get rid of the phasing effect - the delay became obvious, unless if I was hearing things. Those tracks aren't meant to be combined. Technically, Original_1 contains the exact content as Original_2, except Original_1 has more instruments, but is lower quality (I downgraded both to MP3 for smaller file size). Since Original_2 is higher quality, I would to like to "add it" to Original_1 without phasing.

Edit: I will try DC and see what happens. Thanks.

[EmpireStrikesBack198]

I applied DC Offset Correction (called "Repair DC Offset" in Audition) on one track and it didn't fix the phasing issue.

[hello_hello]

It might be unfixable if the two tracks aren't exactly the same.... or if at least the audio they have in common isn't exactly the same. For example if the speed of one track varies very slightly compared to the other for some reason, they might sync for a bit then go out of sync and the phasing effect will come and go as a result. And thinking about it.... if the quality of the two aren't the same that might cause it too. Once again it's not something I've really had to think about before, but if you combine two identical sounds they should simply combine to produce the same sound, only a little louder, but if the frequency response of each is different it might cause them to "interfere with each other" a little and produce a phasing effect. I'm not really sure but maybe that's what's happening.

[netmask56]

From your description it sounds like one track version is out of phase to the other, so if you add them to each other you get major cancellation especially as both tracks are nearly identical. Adding them together but with one track (either doesn't matter) phase inverted they should then add together without producing a tinny sound. In the world of stereo as an example you have L + C and R + C so a mono mix of the 3 elements is represented as L+2C+R if you invert or phase reverse one channel you get L-R so called vocal cancellation. So what I hear is classic cancellation.

PS Time offsetting won't fix the basic problem it will only add phasing effects. In Adobe Audition you can reverse the phase of one channel just as you can in any real world audio mixer. Check out your program to see if it has this facility.

This is what I did

 C:\Users\Netmask\Desktop\testmusid.wma
Format                                   : Windows Media
File size                                : 118 KiB
Duration                                 : 6s 450ms
Overall bit rate mode                    : Constant
Overall bit rate                         : 150 Kbps
Maximum Overall bit rate                 : 129 Kbps
Encoded date                             : UTC 2013-11-06 23:21:24.665
Writing application                      : Audition 3.0.7283.0

Audio
ID                                       : 1
Format                                   : WMA
Format profile                           : Pro
Codec ID                                 : 162
Codec ID/Info                            : Windows Media Audio
Description of the codec                 : Windows Media Audio 10 Professional - 128 kbps, 48 kHz, 5.1 channel 16 bit 1-pass CBR
Duration                                 : 6s 450ms
Bit rate mode                            : Constant
Bit rate                                 : 128 Kbps
Channel(s)                               : 6 channels
Sampling rate                            : 48.0 KHz
Bit depth                                : 16 bits
Stream size                              : 101 KiB (85%)
Language                                 : English (US)
Fold6To2Channels3                        : -558413,-2147483648,-2147483648,-558413,-755021,-755021,-1344845,-1344845,-755021,-2147483648,-2147483648,-755021

[hello_hello]

If testmudis.wma is the two tracks combined after one has been phase reversed then it sounds like a lot has been cancelled out to me.

[EmpireStrikesBack198]

If testmudis.wma is the two tracks combined after one has been phase reversed then it sounds like a lot has been cancelled out to me.

Also, the it sounds tinny on my side. I don't have my headphones with me right now, so it could be the speakers. I will try reversing the phase and see what happens. Thanks.

[hello_hello]

I doubt it'll work but you never know. If the two were identical then they'd cancel each other out, but if they're just different enough to cause a phasing effect before phase reversing one of them, then after phase reversing one they'll probably be just different enough to cancel much of each other out while what remains still sounds "tinny".

IMO the only effective solution would be not to combine them while you're combining them, which means swapping the left/right channels of one before you do, but I know that's not what you're trying to achieve. The only reason that seems to work is because each track is more like two mono tracks.... one in the left speaker and the other in the right..... and not so much "stereo" where the left and right channels also have a lot in common. If they were more "stereo" and less "dual mono" then even swapping the channels of one before combining them probably wouldn't help. If I remember correctly one version does have a percussive sound in both channels but the other doesn't have it at all, so it can't interfere with itself when combining the two.
That's why I didn't mind the "channel swap" method because it produced more of a stereo effect than a dual mono one (due to the phase/frequency response differences between tracks which causes problems when simply combining them), but each to their own....
Admittedly I'm only listening to a small sample though, so even swapping the channels of one track mightn't work for the full version of they contain more stereo sounds in common later on.

[jagabo]

What's happening is this: only one track has bass. The mids and highs are common to both. So when you mix the two together the base is reduced by half compared to the mids and highs. So use an equalizer to increase the bass.

[hello_hello]

I'm sure it's a phase problem. I'm hearing two tracks, one which contains bass instruments in the recording and one which doesn't. Theoretically the bass instruments can't produce phasing problems if they only exist in one track, although it does sound a little like they are, but I think it's deceiving because the non-bass stuff the two tracks do have in common is causing a phasing effect which, for want of a better description, kind of "sits on top" of everything.

Without having tried it, I tend to think simply increasing the bass frequencies of the track without any bass instruments won't help, and might make it worse, because it's the stuff the two tracks have in common which is phasing and not the stuff they don't.
If it was just a matter of relative volumes, then yes after combining the two the bass instruments probably would sound softer relative to the wind instruments etc which exist in both tracks, but it sounds to me like the non-bass instruments are "phasing" when the tracks are combined.

[netmask56]

If testmudis.wma is the two tracks combined after one has been phase reversed then it sounds like a lot has been cancelled out to me.

The test I did was to wmapro 5.1 - sounds ok on my studio system - 5 x 15" Tannoys + subs. So my guess is you were just listening to one track. I'll do a straight stereo mix

[netmask56]

Listen to this one - phase reversed Tk1 and did a stereo mixdown. Sub bass is there. When you mix down without the phase reversal ie as is low bass is lacking.

[EmpireStrikesBack198]

That newest sample doesn't sound bad at all. I can easily add the sub bass with another program, I think. All you did is reversed the phase on one track and does it matter which one? I tried to reverse the phase using Audition, but I still got the phasing issue.

The two tracks have the exact same stereo mix, except one is much lower quality in its original WAV form. The bass instruments are mixed in the center after 5-6 second point. I thinking I should have posted samples in the middle of the track to better display that... if that matters.

[hello_hello]

Well I might need to stand corrected regarding my whole "phase reversal will make it worse" stance, as the second sample doesn't sound too bad, although I'll have to have another listen later when I can crank it up without bothering anyone.

One question though.... which is where I'm confused..... if testmusid.wma and PhaseReversemixdown.wav are both the two tracks combined after one has been phase reversed, why do they sound so different? I guess I'm missing something obvious here or I'm having a miscommunication.

Oh.... I get it.... testmusid.wma isn't a stereo track. I only just noticed. Anyway....

Compare PhaseReversemixdown.wav to the sample in post 2 where I combined the two after reversing the left/right channels of one. When you compare them, PhaseReversemixdown.wav sounds more "staccato" to me so there does seem to be some sort of phase cancellation going on, although admittedly it doesn't sound too bad. I don't quite understand why it lost so much bass though compared to the sample in post 2.
Yep, after another listen I think the sample in post 2 sounds much better and makes the phase cancellation in PhaseReversemixdown.wav obvious by comparison, but unfortunately it's not the result EmpireStrikesBack198 is after.

[EmpireStrikesBack198]

I've been experimenting with different methods the past few days and reversing the phase of one channel is the best option and most faithful. I wanted to thank you both for taking the time to help me figure this out!