Match waveform pattern and apply it to another wave form?

[EmpireStrikesBack198]

I wondering if it is possible to match the patterns of one waveform and apply it to another? For example...





Is there is a way to make L2 look like L1? Is there a way to analyze the pattern and levels of L1 and apply that same pattern to L2? Thank you.

[jagabo]

Yes, copy L2 to L1.

Or do you mean you want to normalize the volume of L2 to match L1? Open the audio in an audio editor and adjust the gain.

[EmpireStrikesBack198]

Yes, I guess what I mean is normalize L2 to match L1... is that possible?

[netmask56]

It would really help if you let us know WHY you want to do this. There could be many reasons you have that could have vastly different solutions.

It is normal in stereo recordings to have differences between left and right otherwise the recording would have little or no separation.

Sometimes the left right balance as perceived by the listener may be lop sided. If this is the case you could select say R1 and raise the level by 2 or 3db - you can do this with range of software from Audition to Goldwave etc... or another approach as others have suggested is to normalise both channels to their max without causing clipping but this does not always address the problem of aural imbalance .

So lets us know what the actual problem is

[EmpireStrikesBack198]

Well, if you look at the bottom picture, you will notice that that waveform is in mono, while the above is in stereo. I want to take the waveform pattern of L1 and match it with L2. And I want to take the waveform pattern of R1 and match with R2, so that the left and right channels of the bottom picture look identical to the top picture (or as close as possible). This is sort of an experiment of mine to make the mono track sound more like stereo. Matching the waveforms is the first step of making the mono track sound like stereo, of course.

I know this may sound confusing and it may not provide me with satisfactory results, but this is something I want to experiment with and see where it goes. The first step is the match the waveforms, if possible.

[edDV]

Audacity

Cut, paste and filter.

[jagabo]

Well, if you look at the bottom picture, you will notice that that waveform is in mono, while the above is in stereo. I want to take the waveform pattern of L1 and match it with L2. And I want to take the waveform pattern of R1 and match with R2, so that the left and right channels of the bottom picture look identical to the top picture (or as close as possible). This is sort of an experiment of mine to make the mono track sound more like stereo. Matching the waveforms is the first step of making the mono track sound like stereo, of course.

Why not just copy the top waveform to the bottom and boost the amplitude (volume)?

[davexnet]

See this forum for mono > stereo enthusiasts
http://bsnpubs.websitetoolbox.com/?forum=6491

[EmpireStrikesBack198]

Audacity

Cut, paste and filter.

I am not sure what you mean here. Cut and paste what?

Why not just copy the top waveform to the bottom and boost the amplitude (volume)?

You mean copy L1 and R2 to L2 and R2? If so, I am not looking to mix L1 and R1 with L2 and R2. What I am trying to do it alter to the waveform L2 and R2 to match the waveform of L1 and R2, so that they look similiar. Is that possbile?

Thanks for the replies!

[Cornucopia]

Why do you think they need to look similar? What are you trying to do? Once you think they are similar, what is your ultimate goal?

I ask this, because when you say "look" while referring to something audio, leaves LOTS of room for subjective error. You could be trying to optimize the dynamic range, normalize to a reference, applying one envelope to modulate another, apply formant processing, copying tracks, mixing tracks, replacing tracks...lots of choices here.

Be more specific with your sources, your equipment & methods, your level of audio understanding and your needs.

Scott

[Noahtuck]

@ Cornucopia

This is what he wrote earlier,

Well, if you look at the bottom picture, you will notice that that waveform is in mono, while the above is in stereo. I want to take the waveform pattern of L1 and match it with L2. And I want to take the waveform pattern of R1 and match with R2, so that the left and right channels of the bottom picture look identical to the top picture (or as close as possible). This is sort of an experiment of mine to make the mono track sound more like stereo. Matching the waveforms is the first step of making the mono track sound like stereo, of course.

I don't think Empire quite understands the difference between true or real "stereo" & "mono"

[Cornucopia]

Well, if you look at the bottom picture, you will notice that that waveform is in mono, while the above is in stereo. I want to take the waveform pattern of L1 and match it with L2. And I want to take the waveform pattern of R1 and match with R2, so that the left and right channels of the bottom picture look identical to the top picture (or as close as possible). This is sort of an experiment of mine to make the mono track sound more like stereo. Matching the waveforms is the first step of making the mono track sound like stereo, of course.

I know this may sound confusing and it may not provide me with satisfactory results, but this is something I want to experiment with and see where it goes. The first step is the match the waveforms, if possible.

OK, I think I have a better idea what you're trying to do, and I can tell you you're going about it the wrong way. Those "waveforms" you see are ZOOMED WAY OUT. If you were to zoom in, you would see a single sinusoidal wave. Applying one sinusoidal wave to another MODULATES the 2nd waveform, either by amplitude, frequency or phase (or a few other more esoteric ones). Amplitude modulation is the simplest of these, and according to what you've mentioned, probably what you were thinking of. But AM produces artifacts, specifically sum+difference harmonics, when given more than a single tone as modulator (and you wanted a whole complex soundfile). You DO NOT want that to happen with your soundfile if you want it to still sound ungarbled.

You MIGHT want to be applying a VERY GENERAL ENVELOPE of the 1st sound onto the 2nd sound, but that wouldn't make sense if they weren't programmatically and, more specificly, positionally related.

Creating Stereo from Mono with any amount of control involves ALOT more than changing an envelope. We're talking something like:
1. Isolation of multiple "individual voices" through filtering & boolean matrixing
2. Synthesis of extrapolated full voices from the filtered voices
3. Posiitioning of the extrapolated full voices
4. Ambience synthesis & admixing

Otherwise, it's going to sound like FAKED MUSH.

Scott

[EmpireStrikesBack198]

OK, I think I have a better idea what you're trying to do, and I can tell you you're going about it the wrong way. Those "waveforms" you see are ZOOMED WAY OUT. If you were to zoom in, you would see a single sinusoidal wave. Applying one sinusoidal wave to another MODULATES the 2nd waveform, either by amplitude, frequency or phase (or a few other more esoteric ones). Amplitude modulation is the simplest of these, and according to what you've mentioned, probably what you were thinking of. But AM produces artifacts, specifically sum+difference harmonics, when given more than a single tone as modulator (and you wanted a whole complex soundfile). You DO NOT want that to happen with your soundfile if you want it to still sound ungarbled.

You MIGHT want to be applying a VERY GENERAL ENVELOPE of the 1st sound onto the 2nd sound, but that wouldn't make sense if they weren't programmatically and, more specificly, positionally related.

Creating Stereo from Mono with any amount of control involves ALOT more than changing an envelope. We're talking something like:
1. Isolation of multiple "individual voices" through filtering & boolean matrixing
2. Synthesis of extrapolated full voices from the filtered voices
3. Posiitioning of the extrapolated full voices
4. Ambience synthesis & admixing

Otherwise, it's going to sound like FAKED MUSH.

Scott

Thanks for the thorough explanation, Scott. I hope I understand what you're trying to tell me here... So, If I were able to modulate L2 with L1, then the track would sound like mush?

It sounds like I am way in over my head on this.

How about this .... Is there a way to match the volume levels of L1 with L2?

[jagabo]

It sounds like I am way in over my head on this.

Yes.

If you make the L2 waveform match L1 exactly all you've done is made L2 into L1. You might as well just keep L1. If you mean you want to make L2 look like L1 at some macro scale, all you've done is modulate the volume of L2 so that the graph of the waveform at the chosen scale looks similar. This will give you some weird volume variations.

How about this .... Is there a way to match the volume levels of L1 with L2?

What exactly to do you mean by that? Do you want the peak volumes to match? The overall RMS volume? Do you want the volumes to match over some arbitrary temporal window (ie, "ride the pot")? The first two are easily achieved by adjusting the gain (volume) in any audio editor.

What exactly is it you're trying to do? Maybe you have a poor quality stereo recording and a high quality mono recording and you want to make the mono recording into a stereo recording using the low quality stereo recording as some sort of guide?

[Cornucopia]

I'll give you some examples of what I was talking about later tonight (once I'm off work and have gotten the kids off the computers - no small matter ).

Scott

[EmpireStrikesBack198]

What exactly to do you mean by that? Do you want the peak volumes to match? The overall RMS volume? Do you want the volumes to match over some arbitrary temporal window (ie, "ride the pot")? The first two are easily achieved by adjusting the gain (volume) in any audio editor.

I think what I mean is match the overall RMS volume... but when you adjust the gain, do you have to do that manually or is there a way to do it automatically?

What exactly is it you're trying to do? Maybe you have a poor quality stereo recording and a high quality mono recording and you want to make the mono recording into a stereo recording using the low quality stereo recording as some sort of guide?

I do have two recordings... one is mono and the other is stereo. The mono recording is a much higher quality sound quality than the stereo recording. The track was originally recorded in stereo, but is very poor quality and I am unable to obtain a higher quality recording. I do have the same track in mono and quality is excellent. Even though the stereo recording is poor quality, it is still in stereo, which is why I want to use the stereo recording as a guide for the mono recording.

I'll give you some examples of what I was talking about later tonight (once I'm off work and have gotten the kids off the computers - no small matter ).

Thanks for taking the time to do that. That should be very interesting to hear.

[Cornucopia]

I do have two recordings... one is mono and the other is stereo. The mono recording is a much higher quality sound quality than the stereo recording. The track was originally recorded in stereo, but is very poor quality and I am unable to obtain a higher quality recording. I do have the same track in mono and quality is excellent. Even though the stereo recording is poor quality, it is still in stereo, which is why I want to use the stereo recording as a guide for the mono recording.

Aaahh, so the 2 recordings are SUPPOSED to be of the same program/clip. That clears a few things up.

Ok, it looks like you want an Automatic RMS Volume Envelope adjustment, after normalizing. I'll try to show that example later on...
The variable there is the slew rate (aka rate of change of the volume envelope, relative to the waveform). If it's too short, you'll get some of those cross-modulation artifacts, too long and it won't track the program changes thoroughly enough (at that extreme, it would be a single gain adjustment).

Scott

[Cornucopia]

I don't know if you've ever heard of ADR (Automatic Dialog Replacement). I've used this before on videos & movies.
Well, there are apps & plugins that can do what I was suggesting, in a very straightforward manner. For example, the VocAlign plugin from SynchroArts (used in ProTools) does EXACTLY that kind of Volume Envelope adjustment (along with Micro-time-shuffling, time-stretching and other things). Of course, that is optimized for Dialogue, not Music.
But you MIGHT be able to coax out of it what you were looking for.
And if you don't have access to a plugin, I'll map out how you might be able to do it with (let me guess...) Cooledit/Audition. (The waveforms you displayed look like they came from that app).

Scott

[jagabo]

I do have two recordings... one is mono and the other is stereo. The mono recording is a much higher quality sound quality than the stereo recording... I want to use the stereo recording as a guide for the mono recording (to make a high quality stereo recording).

Your general idea is sound. You have a stereo recording with left and right channels (L, R) and a mono recording (M) that is essentially L+R (scaled in this case but let's ignore that for simplicity). In theory you can subtract one of the stereo channels to recreate the other stereo channel: R=M-L, or L=M-R. But since the stereo channels are compromised, you will still have the noise/junk if use this simple mathematical approach. And, of course, the tracks have to be perfectly temporally aligned. If you want to try it you can use a stereo-to-mono filter with the mono channel and either the left or right channel inverted.

Cornucopia's approach should work better but I doubt it will be good enough to make it worthwhile except in simple areas where single sounds appear in one location at a time.

[EmpireStrikesBack198]

I don't know if you've ever heard of ADR (Automatic Dialog Replacement). I've used this before on videos & movies.
Well, there are apps & plugins that can do what I was suggesting, in a very straightforward manner. For example, the VocAlign plugin from SynchroArts (used in ProTools) does EXACTLY that kind of Volume Envelope adjustment (along with Micro-time-shuffling, time-stretching and other things). Of course, that is optimized for Dialogue, not Music.
But you MIGHT be able to coax out of it what you were looking for.
And if you don't have access to a plugin, I'll map out how you might be able to do it with (let me guess...) Cooledit/Audition. (The waveforms you displayed look like they came from that app).

Scott

I apologize for not making that more clear that those two waveforms were of the exact same piece of music, except that one is stereo and the other is mono. I've never heard of that plug-in (VocAlign). Is it free? You're right, I am using Audition 3.

[EmpireStrikesBack198]

Your general idea is sound. You have a stereo recording with left and right channels (L, R) and a mono recording (M) that is essentially L+R (scaled in this case but let's ignore that for simplicity). In theory you can subtract one of the stereo channels to recreate the other stereo channel: R=M-L, or L=M-R. But since the stereo channels are compromised, you will still have the noise/junk if use this simple mathematical approach. And, of course, the tracks have to be perfectly temporally aligned. If you want to try it you can use a stereo-to-mono filter with the mono channel and either the left or right channel inverted.

Cornucopia's approach should work better but I doubt it will be good enough to make it worthwhile except in simple areas where single sounds appear in one location at a time.

I am not sure how to attempt this method, but I am willing to try anything. No problem with aligning the tracks perfectly, though. I am not sure how to use a stereo-to-mono filter.

[Cornucopia]

Here's some of those examples:

1. You've got a piece of music



2. You take a LowFrequencyOscillating Tone (sine wave), say 4Hz (Plain View & Zoomed-In)



3. If you Amplitude Modulate it at 50% it's like this



4. But if it's at 100%, it's like this (BTW, AM using LFOs is also known as "Tremolo", whereas Freq.M. is known as "Vibrato")



5. Now, you wanted to take complex soundfile...



6. And do the same thing, which ends up like this



*********

So, what you end up with is MUSH. (Note that I slightly cheated and gave you an output what was Convolved/ImpulseModelled, for expediency)


Now, trying to do this requires specific tools. Unfortunately, Audition doesn't have this in it's standard palette. I can do it in ProTools by using a Gain plugin that can take its control input from a sidechain instead of a standard LFO, but even that takes a little setting up to get right. You sometimes see these "Envelope Follower" plugins married to LP or HP filters to be used on instruments as "auto-wahwah".

The Vocalign Plugin I was referring to isn't cheap (minimum $325). There might be some other plugins out there that could do the trick for you, though (not sure if they're all workable from w/in Audition). Still, takes lots of trial & error.

I'd start with jagabo's suggestion.

Match the timing of the 2 clips-to the sample, if possible.
Then, do a mix of Mono (L+R) with each channel of the L+R stereo track, but with the Mono OUT-OF-PHASE (inverted amplitude). Adjust the relative level until the L of the Mono cancels out the L of the stereo, and you're left with just the R. Do the same thing for the other channel. Unfortunately, it won't work perfectly (or maybe even acceptably) because as you said, they are REALLY matched in quality.

You also could try a few of the "Stereo Imagery" preset effect filters in Audition. They're good in a few instances.

Scott

[EmpireStrikesBack198]

So, what you end up with is MUSH. (Note that I slightly cheated and gave you an output what was Convolved/ImpulseModelled, for expediency)

Yes, that sounds horrid... for a positive listening experience that is. It seems like it is too much work to bother even attempting. Thanks for the the visuals and sounds... that really helped me better understand what you were talking about in previous posts.

I'd start with jagabo's suggestion.

Match the timing of the 2 clips-to the sample, if possible.
Then, do a mix of Mono (L+R) with each channel of the L+R stereo track, but with the Mono OUT-OF-PHASE (inverted amplitude). Adjust the relative level until the L of the Mono cancels out the L of the stereo, and you're left with just the R. Do the same thing for the other channel. Unfortunately, it won't work perfectly (or maybe even acceptably) because as you said, they are REALLY matched in quality.

What are you referring to when you say "R" in this statement: "Adjust the relative level until the L of the Mono cancels out the L of the stereo, and you're left with just the R."

Also, how do I know when the L inverted mono cancels out the L stereo? I mixed the inverted L mono with the L stereo at different levels, but they didn't sound any different from each other.

[chowmein]

Have you thought of just mixing the single mono channel in with the poorer-quality stereo track? It's not a perfect solution, but you may be able to achieve a better sound quality while still retaining some of that stereo "spread".

[EmpireStrikesBack198]

Have you thought of just mixing the single mono channel in with the poorer-quality stereo track? It's not a perfect solution, but you may be able to achieve a better sound quality while still retaining some of that stereo "spread".

Yes, I have. That is why I wanted to "match" the mono waveform with the stereo waveform, so the mono wouldn't "override" the stereo, ruining the sound of the stereo. Unless if I kept the mono volume below the volume of the stereo track.

[jagabo]

That is why I wanted to "match" the mono waveform with the stereo waveform, so the mono wouldn't "override" the stereo, ruining the sound of the stereo. Unless if I kept the mono volume below the volume of the stereo track.

If you mix the two in equal proportions you cut the stereo separation in half and keep half the problems of the stereo track. If you mix a small amount of the stereo track you get a small amount of separation and only get a small amount of the problems of the stereo track. If you mix a large portion of the stereo track you get a large stereo separation but also get most of the problems of the stereo track. So you don't necessarily want to mix them in equal portions. "Portion" here means the volume of the tracks, the height of the waveforms. So you don't necessarily want to match the waveforms. You want to look for a good compromise.

You also need to make sure the phase of the two signals match, and the phase shift is as small as possible.

Another problem with trying to mix the tracks like this is that a large part of the stereo perception is carried in the high frequencies. Your low quality stereo tracks probably don't have a lot of high frequencies. The mono track probably does. But they will be added in equal proportions to both the left and right channels with a simple mix -- mostly eliminating the purpose of the mix.

I started with a high quality stereo track. I merged the two channels and amplified a bit to make a high quality mono track. I also made a separate low quality stereo track by adding a little white noise and encoding as MP3 with a low bitrate (64 kbps, joint stereo). Here are the two waveforms:



Zooming into the waveform we can see that the MP3 encoding has caused a 51 ms phase shift (delay) in the low quality stereo track:



So I shifted the the low quality stereo track left by 51 ms:



The phase shift has been corrected. You can also see that the phase of the two signals matches. When one signal goes "up" so does the other signal, not the other way around. You don't want the signals to be phase inverted:

[EmpireStrikesBack198]

If you mix the two in equal proportions you cut the stereo separation in half and keep half the problems of the stereo track. If you mix a small amount of the stereo track you get a small amount of separation and only get a small amount of the problems of the stereo track. If you mix a large portion of the stereo track you get a large stereo separation but also get most of the problems of the stereo track. So you don't necessarily want to mix them in equal portions. "Portion" here means the volume of the tracks, the height of the waveforms. So you don't necessarily want to match the waveforms. You want to look for a good compromise.

So, when I mix the stereo and mono together, the mono portion will probably be above 50% and the stereo portion below 50%. Should I spread the mono track so that when I mix the stereo and mono together it won't sound too center heavy?

[smartel]

have you ever seen this : http://www.har-bal.com/

i don't know if i'm off topic !?

[jagabo]

So, when I mix the stereo and mono together, the mono portion will probably be above 50% and the stereo portion below 50%.

Using the tracks pictured in your first post as-is? Yes.

Should I spread the mono track so that when I mix the stereo and mono together it won't sound too center heavy?

If you can spread the mono track why bother mixing in a low quality stereo track?

[EmpireStrikesBack198]

have you ever seen this : http://www.har-bal.com/

i don't know if i'm off topic !?

Hmm, seems like this would be a very useful tool. However, it is well above my price range and they don't offer a trial version to try it out. Thanks for the sharing that link, though.

If you can spread the mono track why bother mixing in a low quality stereo track?

True. I was concerned about it being to center heavy since the stereo track has a center portion as well. Unless if I extract the center portion from the stereo track. I will try it out and see how it sounds.