Interpreting audio waveforms

[Jimbo]

Anyone care to explain how the top half of an audio waveform relates to the bottom half?

I know that the waveform shows the signal over time, but what does the direction (up/down) of the waveform mean? I'm guessing that it's frequency related with the top being the amplitude of the signal going higher in the freq spectrum, and bottom being the amplitude of the signal going lower in the freq spectrum.

Thanks,

- Jim

[manning1]

if you look at a waveform in an audio editor.
you will see the waveform revolve around a center horizontal line.
(i suggest you google for the old cool edit 96 editor if you dont have one.
its pretty good and shows you all sorts of measurements on a waveform down to
sample values. it will help you learn as the docn is pretty good...but 16 bit only.).
the horizontal line is where amplitude is lowest.
The tops and bottom of the waveform are where the sample amplitudes
are highest. then they gradually get lower in amplitude crossing the zero line before rising again. ie...AC waveform.
therefore a rule is when cutting audio (to move it around) you should cut at zero crossings (the center line). thus no pops or clicks. you shouldnt cut at high amplitudes. hope this helps. cool edit 96 also includes a frequency analyser so you can hilite some audio and do freq analysis.

[Jimbo]

Originally Posted by manning1

The tops and bottom of the waveform are where the sample amplitudes
are highest. then they gradually get lower in amplitude crossing the zero line before rising again. ie...AC waveform.

Ah, AC waveform...got it! Great explaination Manning, thanks!

Now, here's a followup question, and the reason I'm asking about the waveforms in the first place.

I was testing-out a compressor that a friend gave me (a Demeter VTCL-2), and noticed that even on mild settings it seems to suck the bottom-end out of vocals. I then looked at the waveform, and noticed that the bottom half of the waveform was compressed (less amplitude), but the top half stayed the same. What's going on here?

The compressor is unbalanced, and the rest of the signal chain is balanced. Could this be the cause, and, if so, what is the workaround?

Thanks,

- Jim

[tINY]

If the bottom half of the waveform is the only one getting compressed, then the gain reduction is "asymetrical". This is the same thing that class A tube amplifiers tend to do.

It ususally is considered a good thing when overdriving an amplifier, as asymetrical clipping gives you more even-order harminics that sound more musical. When the tops and bottoms of the waveform are both truncated, you get more odd order harmonics that tend to sound "harsh".

But, a compressor should simply reduce the amplitude evenly everywhere when it is working.

I'm guessing that you are overdriving the poor compressor......

(By the way, compressors tend to suck the life out of things - that's what they do)




-tINY

[Jimbo]

Originally Posted by tINY

I'm guessing that you are overdriving the poor compressor......

I'm applying very light compression on the way in with this unit. I don't think that I'm overdriving it since it has an overload indicator that never even flickers.

Originally Posted by tINY

(By the way, compressors tend to suck the life out of things - that's what they do)

Well, yes, if they are used that way, which is not the way I use them. However, I'm trying to understand why the compressor seems to be affecting the bottom end, and not the top-end, and if that relates to the odd waveform that is generated.

It's a vacuum tube opto-compressor. I was adding about 10db of make-up gain for shits and giggles. I think I'll try the unit with no compression, only adding makeup gain to see if that's causing it.

Thanks.

[ulysses]

The vertical axis in the waveform view on your DAW represents positive and negative voltage swings in the analog electrical signal. That electrical signal in turn is an analog of the waves of compression and rarefaction of the air between the acoustic source and the transducer (microphone). So, technically, "up" in the DAW window represents a string or membrane (drum head, guitar string, etc) moving toward you while "down" represents it moving away from you. It has nothing to do with frequency - this is the amplitude component of the waveform. The distance between those peaks would be the period of the waveform, and the reciprocal of that is the frequency.
When you see one half of the waveform having greater amplitude than the other, that is a visual representation of the effect of 2nd-order harmonic distortion. Like tINY said, this is the kind of artifact that sometimes causes the analog to sound better than the original. There are lots of things within the compressor circuit that can cause it. Whether the device has balanced or unbalanced interfacing has nothing to do with it, but push-pull circuits do tend to cancel 2nd-order distortion.

To clarify what Tiny said - It isn't that Class A amplifiers exhibit this behavior. It's only single-ended amplifiers. Since single-ended audio amps have to be Class A to function correctly, the association is erroneously made. Push-pull amplifiers operating in Class A don't exhibit this behavior - they cancel any assymetry of each amplifier half just as class B amplifiers do.

[tINY]

Fair enough, Ulysses.

What may be going on is that he has a single-ended class B amplifier on his hands....

Sounds like a DC offset has drifted way off or one of the rails is being loaded down excessively. Opto compressors tend to work slowly, so I would suspect an amplifier stage of causing this behavior. On VCA (log-log) based compressors, the actual compressor could cause this.





-tINY

[manning1]

jimbo...
i bow to tiny and ulysses. really listen to these guys...great info chaps.
so you worked at intel eh tiny ??
are you retired from the computer industry like me ??

[Jimbo]

Ulysses: Thanks for the "better-than-textbook" explanation. All three of those paragraphs were extremely enlightening.

tINY: You were spot-on. I was overdriving the compressor. This compressor isn't like others I have used. It's a limiter with a soft knee. There is no ratio, rather an "Input Sensitivity" and "Gain Reduction" pots. The labelling on the unit is somewhat confusing, and the manual is only one page -- you can find it on the Demeter website if you interested -- and the manual uses a different nomenclature than what's actually labelled on the compressor. It was given to me by a friend, and I've only used it a few times. I've always just fooled around with it until it sounded good.

---
So, I did a test this evening. I ran vocals through the compressor at 5 different settings, using the makeup-gain pot to normalize the output to -6dBu

1. "Bypass" - Bypassing the compressor resulted in no shift in the resulting waveform.
2. 0dB Gain Reduction - When the gain-reduction pot was set to zero, there was no shift in the resulting waveform. However, the amplitude of the waveform was compressed, slightly -- Interesting!
3. -10dB Gain Reduction - With the gain-reduction pot set to -10, there was a noticible shift in the resulting waveform. However, the amplitude did not look any more compressed than at "0dB Gain Reduction".
4. -20dB Gain Reduction - With the gain-reduction pot set to -20, there was the same shift in the resulting waveform as the "-10db" setting. The amplitude did not look any more compressed than at "0dB Gain Reduction".
5. -30dB Gain Reduction - Results were the same overall as with "-20dB".

Of course, as the gain-reduction was increased, there was increased compression of the signal, but the resulting waveform was fascinating.
The fact that even with zero gain reduction, the fact that the signal was going through the compressor circuit, AND there was a waveform shift confirms what Ulysses said about the second-order harmonics causing shift.

This compressor -- when used correctly -- definitely has a soft touch. Now that I understand it, I expect some good things to happen when tracking vocals or sloppy fiddle players.

I perform, compose, and record acoustic music (bluegrass/folk/americana). The recording part has been just a hobby, but I recently took it up a notch when I decided to get dirty with electronics and start building gear. The geeky part of the biz is really interesting for me -- However, I have no training in electronics or mechanical engineering, so this is new territory for me.

I really appreciate everyone's help with this. I'm going to stay tuned to this section of the forum.

Thanks!

- Jim

[tINY]

Originally Posted by manning1

are you retired from the computer industry like me ??

I am not retired.



-tINY

[moogus]

But does it sound bad to you? Some of my favourite gear has a very asymetrical output. Admittedly its very 'soundy' gear that I wouldnt record, say, vocals with often.

You dont have access to a CRO and a signal generator do you? If you do, see if its asymetrical at all freqs. Test 1k, 100hz and 20hz and see if it changes.


M

[Jimbo]

Originally Posted by moogus

But does it sound bad to you? Some of my favourite gear has a very asymetrical output. Admittedly its very 'soundy' gear that I wouldnt record, say, vocals with often.

You dont have access to a CRO and a signal generator do you? If you do, see if its asymetrical at all freqs. Test 1k, 100hz and 20hz and see if it changes.

M

As tINY suggested, I was overdriving the compressor, and it was sounding bad. Now that I've figured-out how to use the dang thing, it sounds quite nice. Pretty transparent too at moderate levels of compression.

It's actually the type of compressor I've been looking to use while tracking. The soft-knee limiting will prevent vocals and ruthless fiddles from accidentally clipping the converters, while smoothly leveling some of the dynamics. It can be run in dual mono or linked stereo mode as well. I don't mix outside the box, so this is just for tracking. Best of all, it's free

I'm going to ask our EE at work if he has a signal generator. Our electronics lab has all kinds of "stuff" that I'm unfamiliar with.

Thanks,

- Jim

[emerson77]

hey guys! thanks for this thread - i'm having an almost similar problem, but i have one variable that i was wondering if you could clear up.

my mixes from PT via an 888/24 to my ART Pro VLA (my favorite piece of cheap gear) created the same asymmetric waveform, even at very minor gain reduction.

the only way i was able to get it 'look' normal was to reduce the output of my master fader in PT by -12dB and then make up for it with the gain on the ART.

do you think this is a problem the the ART or the calibration or my PT rig?

and

do you think there is anything wrong with sending so little output from PT to an piece of outboard gear?

thanks a million!
matt
http://www.emerson777.com
MySpace.com - emerson77 - SEATTLE, Washington - Indie / Ambient / Alternative - www.myspace.com/emerson77prod

[ulysses]

It's probably either a lack of headroom or a rather low compression threshold in the ART. But to prove it to yourself, try the exact same signal chain without the ART (just loop through from the PT output to the PT input). I bet you'll find that the PT hardware is just fine.
If the ART Pro VLA is your favorite cheap piece of gear, I'm going out on a limb and guessing you like the way it sounds. If one of the things it does it to make the waveform asymmetrical, then I might further guess you like that as well. If this is the case, then you can quit worrying about what it looks like and get back to thinking about what it sounds like.
The visualization of the waveform offered by the DAW can be a helpful tool. For example, if there's a bunch of DC offset for some reason, that will limit your signal headroom. That would be something to fix. But if the distortion in your favorite analog gear causes the waveform to look different than you expect it to, that isn't really something that needs to be fixed. Remember, your goal is to make pretty sounds, not pretty pictures.

[tINY]

Amen



-tINY

[Radioman]

@Jimbo and Emerson77
But the defect appears equally on both channels 1 and 2 ?
If yes, it must be a schematic signature, and can't be eliminated without a partial redesign of the circuit.
Otherwise you can try to swap the tube pairs (12AX7A - 12BH7A for Demeter while the 12AT7A is only one, 2 x 12AT7 for the ART) and look if the defect move from a channel to the other.
If nothing changes you can take a look for the right voltages of the power supply (+/- 15V an 46V for the ART)

[dwire]

Thank your "Mr. Radioman" I could see what was in the A.R.T. PRO VLA's power supply the last time I was jerking around with one of them, but it is nice to have the manufacturer's schematic to tear it apart and re-think less the magnifying glass and soldering iron. Always best to "tinker" and evaluate in the mind first! BTW, I do have a larger than life schematic of the rest of the circuit around here in the files... I'll add its PSU to the collection of schematics in my files too!

(ART's stuff sure beats a lot of the pencil drawn, over a cocktail stuff I see a lot from the good-ole' days; complete with stuff scratched out and values added in... Kind of neat to see those from giant manufacturers, even the ones now defunct. I always wondered, what sort of "flapper gathering" or "titty-bar" these people came up with these drawings, and ideas in!)

Thanks and best regards,
dwire