Skip to main content

Notice

Please note that most of the software linked on this forum is likely to be safe to use. If you are unsure, feel free to ask in the relevant topics, or send a private message to an administrator or moderator. To help curb the problems of false positives, or in the event that you do find actual malware, you can contribute through the article linked here.
Topic: Best to split/cut audio at the "rest" position of the wave? (Read 771 times) previous topic - next topic
0 Members and 1 Guest are viewing this topic.

Best to split/cut audio at the "rest" position of the wave?

Hi,

For a while I have held this superstition, that... if you're editing/trimming audio, it's best to make the "cut" at a point where the audio is fully "still" in terms of amplitude. This is to avoid the "audio artefact" of an audible click, that is produced (at various loudness-es, depending on severity) when audio is played, that starts at some sort of peak or trough (i think a click might also be produced if audio abruptly ends whilst at a peak or trough, too.

The idea being, if the cut is made where the amplitude is pretty big actually, then you would get a louder click, and so picking a totally "still" position to cut from, eliminates this risk.

I first heard this with some hip hop music that used sampling and looping, and then later was able to confirm the phenomenon (years ago) when I was messing around with similar styles of music making. This can also be heard when using some modern "analogue" monophonic synthesizers that don't have proper procedures for ending a note, (i heard it a lot on my korg monologue). But as of recent times, I am only having to think about this when I am cutting and splitting audio, mostly if i'm making splits in LP rips, or in tape rips, and it's a rule I RELIGIOUSLY follow.

But recently, i got curious to see if this was something other people considered/thought of/knew about on the internet, and I wasn't able to find any discussion of the phenomenon.

I will attach 2 pictures, to show an example of what I mean by, picking a place in the audio that is totally still in terms of amplitude (as in, my current procedure for cutting audio).

The 3rd picture shows a completely separate recording taken off of tape, it was cut by somebody, and it starts at a non-zero position in terms of amplitude, and at the beginning of this track, a noticeable, digital click can be heard as the audio starts.

It should be noted, in my experience, sometimes the amplitude rises "too fast" even from the rest position, so even by following this rule, a click may be audible due to that fast change, but I have not confirmed this in a very long time, and i don't want it to be included as part of my main thesis.


The images are in this google drive folder (there is nothing else in this folder), and are numbered appropriately. If there is a better way to host these, I am willing to make the change too. https://drive.google.com/drive/folders/1XfNm1C4uV-Q9T0YX_5orZ1p9DjoavPcD?usp=sharing

I just want people's ideas on this idea, I can't remember if it's wholly rooted in reason.

Re: Best to split/cut audio at the "rest" position of the wave?

Reply #1
It's called  "zero crossing" and Audacity (and probably other audio editors) has an option of Select At Zero Crossings.  

Another option is a fast fade-in, fade-out, or crossfade (a few milliseconds if you don't want to "hear" the fade).


BTW - A DC Offset can also cause a click where you cut or at the beginning & end.   (A fade will fix that too, or there are other fixes.)

Quote
It should be noted, in my experience, sometimes the amplitude rises "too fast" even from the rest position
With digital audio there's no data between samples so zero-samples are rare and you are unlikely to find an exact zero-sample where you want to cut/splice.    You have to find the the sample just-before or just-after the zero-crossing.

And with Audacity you may have to split the left & right channels and edit them separately because by-default it will select the same left & right sample (time).

Re: Best to split/cut audio at the "rest" position of the wave?

Reply #2
Hi DVDdoug,

thanks for the reply. Huh, I had always been doing this manually (looking for such crossings, where they happen simultaneously in each stereo track).

That said, I have checked out the link you included (thanks!), and it looks like Audacity's built in feature for this is probably gonna replace my current method of doing things tediously and meticulously. I did already know about the fast fading technique, but for some reason didn't think to mention it, (because i dont use it, probably).

DC offsets are pretty easy to spot, to my knowledge, right? it's just where the whole "base" or "centre" of the track is not actually at a zero crossing, hence it's an offset. I don't encounter this very much, I don't even remember what I do in those circumstances, but fading here is probably the easiest solution. I know audacity has it's own DC offset correction tool, but haven't used it, anyway that's a little bit off topic.

Thanks for the reply, I just wanted to make sure I wasn't going crazy... So it's much appreciated.

Re: Best to split/cut audio at the "rest" position of the wave?

Reply #3
Quote
DC offsets are pretty easy to spot, to my knowledge, right?
It depends on how bad it is.  ;)   Sometimes a waveform can be asymmetrical with a higher positive or negative half but zero isn't actually offset.   I think Audacity's offset correction will actually create a DC offset in that case.   A high-pass filter will remove the "long term" DC offset but it can leave a "tick" at the beginning

Quote
it's just where the whole "base" or "centre" of the track is not actually at a zero crossing, hence it's an offset.
Right.  DC means "direct current" and it's "zero Hz".   Of course there is no electrical current in computer file so technically, it's a "zero offset" or a "bias".   There's almost always something in the analog electronics to filter it out so the DC doesn't go to the speakers.   If it does go to the speakers, of course you can't hear zero-Hz and your speakers can't reproduce zero-Hz but you can hear the click when it kicks-in and kicks-out.  Have you ever connected a flashlight battery (DC) to a speaker?

...If the offset is greater than the signal peaks, there will be no zero crossing!  ;)

Quote
I don't encounter this very much
Me neither...  It's usually caused by a defective soundcard.  I think I have a video capture device that puts DC offset in the audio.

Re: Best to split/cut audio at the "rest" position of the wave?

Reply #4
Well, I will have to say, on this occasion I used a combination of my "judge by eye and ear" method, and also some fading where this wasn't possible. I was working with... well I don't actually know what, the accompanying audio from a cartoon's master tape, or at least a broadcast copy. I encountered something that usually only I see on record digitisations, really low inaudible bass "wobble", making the whole wave form look a lot wilder than it should... guess it's called wow&flutter when it comes from tape.

By the way, I am using a "cosine fade" when I fade in or out, this is back from digital music "making" days, where the cosine fade is known to be the most subtle cross-fade tool, except.. crossfading into nothing, of course. I wonder if you have a preference in this regard, or if you do not notice a difference, I know the "studio fade out" is a good looking option often, but i'm not sure it sounds right when the fade is sudden/quick/abrupt.

DC offsets are certainly, at least to me, something that results from a less than ideal recording setup, faulty equipment certainly makes sense to be probably the largest culprit in creating this.

Either way, I will not hold you hostage to this conversation, my concerns and queries have certainly been met, I appreciate you taking the time out of your day to do that. If you don't have the free time to reply, it's all good.

And thanks again!

Re: Best to split/cut audio at the "rest" position of the wave?

Reply #5
Quote
I encountered something that usually only I see on record digitisations, really low inaudible bass "wobble", making the whole wave form look a lot wilder than it should... guess it's called wow&flutter when it comes from tape.
Again, a high-pass filter can take-out subsonic information (or other low-frequency noise). 

Wow & flutter is a speed variation that results in frequency/pitch variations...   You're probably thinking of record warp.  Sometimes you could see the woofer pumping in & out.      Some stereos used to have "rumble filters" and you might still find that on a phono preamp.     I've never actually heard wow or flutter from anything that wasn't badly broken. 

Re: Best to split/cut audio at the "rest" position of the wave?

Reply #6
Hmm, well it looks like i got the term wrong, apologies, I don't often deal with tape... But the sub bass "tremors" that can be seen on the visualiser is definitely a thing, and actually it might be caused by different things for records and tapes. On records, i think it's a specific type of groove damage that registers in the "low" diapason instead of something audible and "clicky". Inconsistent wobbles on tape however, don't make any sense to me, but I have seen that too. Plus on tapes you can have a totally different issue of a constant bass "hum" that creates the same problem, and that can only be dealt with by fades.

Wow&flutter, which I *have* heard a lot.. is the pitch related stuff, as you have put it, and i was wrongly using the term here. I will say this: I know that on properly remastered (from record, not master tape) recordings, done by very high end and respectable studios, even they are not messing with the "sub bass" tremors they come across, so potentially current methods of removing this issue are at least a little destructive to the end product.

Things like kick drums and electric bass after all can pretty much extend down to the lowest bass frequencies. So when that is a consideration, i guess it's a compromise needs to be reached.

I have briefly looked at what rumble filters are (new term for me), on the surface of it, they don't seem to be such a great idea, once again, compromise territory, but I might be wrong. Also, im sure you know, but I am talking about a wobble so low that it's inaudible, and can only be spotted by looking at the waveform, so rumble filters very likely weren't designed to combat this.

Re: Best to split/cut audio at the "rest" position of the wave?

Reply #7
You could just as well cut and splice where the level is the same in both clips, but far away from zero. This should be the place where the overall character of the signal is similar, and the levels also match in both stereo channels at the same time. I sometimes delete a few samples and listen for a thump, then delete some more or undo.

Cutting also works at drum hits or other noise pulses which would mask the click as being part of the program.

The crossfade should probably be a "constant power" crossfade with random usually uncorrelated signal to avoid a level dip in the middle. Basic audio editors usually have a linear crossfade.

I suspect that the zero crossing is not necessarily the place where frequencies of interest are all at zero. And those that are not will make a pop sound. With any phase filtering, such as highpass, the zero will "move". You definitely should apply some gentle highpass to vinyl, but avoid steep filtering and linear phase.

Re: Best to split/cut audio at the "rest" position of the wave?

Reply #8
I figured out on my own as well, that I needed to edit as close to zero as possible. Lately though I've found that I don't notice it if the cut happens up to -25db. I've been using GoldWave, so if the joined waves don't align well enough, I just hand draw them to connect better.  :D
Processed audio in java and python.

 
SimplePortal 1.0.0 RC1 © 2008-2021