Hello, just wanting to write about an unusual waveform that i did want to correct but was unable to.
The content is just voice but the author of the compilation had to return a number of times to finish the work. But by looking at the waveform it appears the microphone he used on two occaisions produces a much different quality of sound and is plainly AC asymmetric.
I had chatted to steve via email who suggested to try a low-pass filter. I did it with default settings and it appeared to work (well, there was minimal offset). I later did it by specifying the cut-off to be beyond the audio range and there was no effect, so this indicates the offset is infact in the upper frequencies.
When looking closely at the sound elsewhere in the compilation it appears the other microphone is also asymmetric.
So the questions are, is this common to all microphones or due to the human voice - i suspect it may even be from having someone’s breath (air pressure) against it . And is it correctable? - at least the major offset? Some solutions would be, to just copy the lower to the upper but better would be an amplification of the upper to meet the lower after first analysing how much to each frequency, and to search to see if the waveform has different standards like in this case.
Human voices are asymmetric — sometimes startlingly so. There is a radio show with a man and a woman tossing it back and forth. I can tell who’s talking just by inspecting the blue waves. She’s more or less even with maybe a slight positive boost, whereas he has a pronounced negative peak bias. All perfectly normal.
Is there a question about the show? Is it badly distorted? If you try to “correct” the blue waves, you may create significant distortion. You already got our attention with the six-hour surveillance recording.
The lead (left edge) dense part in the right-hand bracketed segment does appear to be damaged. I don’t think that’s normal sound. Is that someone turning on a microphone with the sound system running? Some other damage?
I’m still curious what problem we’re solving. If the sound settles down to the exact middle of the range when everybody shuts up, then you have a normal, good quality recording of someone’s voice. If the sound quality shifts from recording to recording, then some other normal problem may be happening like multiple different microphones or different spacing between microphone and performer. That’s why there are still millions of different microphones on the market and people still pay someone to go out and shoot sound.
Some of that you can cure with Effect > Equalization and that, too, is where experienced people work out very well. We can’t tell you “move this slider” to make everything perfect.
If you have a presence (closeness) problem or echoes in the performance, that may be the cue for a reshoot. We can’t solve that with any tools.
+Images show an offset of some kind in parts of the recording, this is what the whole topic is about. If you are disinterested enough to not read then also do not post. If you are a busy person and acknowledging the problem would mean that you would need to clutter your schedule further, dont post. The image (first line of second image towards the end) shows no DC bias (why suggest that?) if there was DC bias i would expect a silence line about 5% off centre.
+Removing high frequencies removes alot of the offset, as shown (suggesting to me that it is likly an effect of the microphone with positive air pressure against it or inherent in the microphone design, having different responses to waveform vacuum than pressure)
+The fact this distortion is not on the other microphone (majority of recording) means that it is a defect in the recording, and i am requesting a tool to remove it. I could send a recording of the same word on both microphones if that would help.
my temporary fix is decreasing the volume of the offset sections, its not the best solution, but if anyone wants to attempt a fix or attempt a filter which does as i suggest i would be happy, otherwise forget it.
Applying DC offset can make an asymmetrical waveform more symmetrical about the zero line.
Steve’s plug-in “tool” for Audacity, which I mentioned in my previous post, applies DC offset dynamically : more-so on the more asymmetrical sections, see “current version” here
[ so it wouldn’t offset a long section of silence ].
it appears the microphone he used on two occaisions produces a much different quality of sound and is plainly AC asymmetric.
Since asymmetric vocal waves are perfectly natural and normal, which microphone is wrong?
I probably would have stopped with: “Can you help make these two performances sound the same?” That is the goal, isn’t it?
I’m guessing nothing so far has helped because they don’t change the harmonic/aural content of the show. They just make the blue waves “look better” by rudely shoving them up and down during the active voice. Since the tool actually produces DC content in the show, it may produce thumping errors in equipment downstream.
You need a software patch that applies different corrections to positive and negative portions of the waveform — and then corrects the resulting DC shifts. That will make the voice sound very different and if your moons and stars line up, corrects the match problem.
Since that’s not a normal sound correction, that might be way down at the bottom of the Nyquist toy box.
There are many reasons why waveforms can be asymmetrical. The really damaging one is DC offset, but that is simple to solve, and does not appear to be present to any significant extent in your screenshots. Most other reasons can be classed as “normal”. Beyond synthesized sounds waveforms are (almost?) never perfectly symmetrical, though examples that exhibit a high degree of asymmetry are relatively rare. Occasionally there can be things going on that are not “normal” (and not DC offset) that may or may not constitute a problem.
Apart from DC offset, asymmetry is more often than not, nothing to worry about, but I’d be interested in taking a very close look at those waveforms to see what is actually going on.
Please post short audio samples (just a few seconds for each) in WAV format (other formats may mess up the waveform). I would prefer that these were totally unprocessed (exactly as recorded) but if you have done anything at all to the audio other than record it, please describe exactly what you did.