How do u find clipped audio?

A straightforward question, possibly a very easy one, but since I'm an idiot and don't know the answer I've decided to make a thread on this

I know stuff like izotope rx exist, but there's a problem with that...

If u run the waveform statistics on rx, it will analyse true peak, sample peak, max/min rms levels, etc. of the audio imported, but if u just lower the clip gain, the readouts will change, the true peak and sample peak levels will change to negative values, as well as the values of the possibly clipped samples.

So the question is, how do one really identify existing clipping in samples? there has to be a way to analyse the audio so that no matter if u lower the clip gain, the readouts will still show that X amount of samples are clipped.

Say for example, that I render a track to 44.1khz / 16bits but I didn't use a limiter, obviously the sample peak value will now be 0dbfs and the true peak higher than that, but if I just lower the clip gain to -4dbfs and analyse the track, the analyser will show me negative values in the sample and true peaks, it won't identify the samples that are already clipped.

Suggestions about what software/vsts to use?

I'm using Reaper btw, thanks a lot for the feedback! x

 

Last edited: Jul 25, 2022

Use Audacity : free and simple

Normalize at 0dbfs and use clipping detection
https://manual.audacityteam.org/man/find_clipping.html

 

You mean hard-clipped? You can see obvious clipping on the waveform, like when peaks have clearly been cropped off. When I master I use both clipping and limiting (several in series and with oversampling enabled), where limiting has more of a soft-clipping effect. But it still has some roundness to the peaks. But I can spot obvious (and hard-handed) hardclipping on the waveform and by the distortion/harmonics it's adding, especially to lower fundamentals/frequencies.

 

Lately I've noticed this new trend with digital distortion (highly desired and intentional in the mix): Distortion in the digital domain has always been seen as evil, but now that's not the case anymore. I see it as an interesting phenomenon, more or less the same step we had in the 70's with analog distortion. Basically, the only valid theorem is: If it sounds good, it is good.

Listen to this: + 14db on the drop! (no limiter)

Ok, you must be under 17yo to enjoy this music, but I like the concept of not setting creative limits.

 

I think that that won't work bc why it has to be normalized to 0bdfs in order to indentify clipped samples. the whole point of my question is to find a way to identify clipped samples in spite of the audio being at a lower volume.

Let's say I take a very very old recording/track and want to know if it has clipped samples. Shouldn't I be able to do that even if I lower its volume before doing the analysis?

 

ya, I get this whole idea. same thing applies with aliasing

 

Yes, I mean hard-clipped distortion, the distortion happening when peaks go above 0dbfs; u are correct when u say that they are obviously noticeable (sometimes not so much, but whatever...), but don't u think it's extremely tedious to go and check every single section of the audio to see if the peaks are flat (distrorted) or not? What if u are working with a 1 hour set instead of a track? Would u waste time doing that and count the distorted peaks one by one or would u like to have an analysis tool that can provide u that information?

Also, a quick question about ur clipping/limiting usage in mastering. do u usually soft clip with a limiter before using a hard clipper? I know that the common thing is the opposite: using a hard clipper then round it off and finish the master with soft clipping using tape or a limiter that provides u with that. Obviously I know it's a case by case situation, but what's ur general approach to that?

 

Last edited: Jul 26, 2022

Virtual Riot is a genius, not all of his music is hardcore dubstep, hes actually quite known for his melodic EDM, and I can't name anyone who would dislike that, tbh.

 

I don't like it, so now you can. My name is Lois Lane, but I'm not the only one, I'm quite sure.

 

@holdup13 Don't worry about that: Lois Lane is an ancient boomer, he is not able to understand the new underground movements of us digitally distorted young people

 

I would use Acon Digital De-Clip2 and get it over with. The audio is already damaged and it will do a decent job as possible. Anyone who wants to say do this all manually, etc.etc can talk to the dev of the plugin. I didn't code it and don't care to hear it. Use it or not.

 

Assuming that the audio is damaged in the first place is what u are doing wrong here. That's the thing that needs to be established before doing the processing, or would u apply a declipper if u don't know if the audio is clipped beforehand? The point is that we need to know if the audio is clipped first to then use declippers, otherwise u will do an unnecesary process that will alter the audio negatively.

 

I'm sure you can like some of his music. It's not all wubs and yoys. But I don't know what kind of music you like.

 

I see your point, but my guesswork logic is that I assume that we are suspicious of the file having distortion in the first place. If there is no digital clipping in the specific file, the de-clipper should do nothing. That's where I thought you were headed. Audacity has a feature which purports to do this. https://manual.audacityteam.org/man/find_clipping.html

 

how much do you hardclip? Because last time I tried it on master it was just sounding bad if more than 0.5 - 1 db of clipping

 

RX declip shows where clips are (before declipping them), and it's threshold can be altered to let/not let certain clipping through.
You can put the declip module on the master or on specific channels or subgroups to further narrow down what is wanted and what not.

 

True peak is in relation to DAC so lowering the volume would mean fixing true peaks. True peaks are not "printed" in the audio, they're a factoring of the digital to analog conversion that introduces intersample peaks such that heard audio has clipping not present in the DAW but present once turned into actual heard audio. If you get a grasp of why you're monitoring true peaks in relation to DAC then you'll understand that by definition lowering the volume can and should eliminate true peaks.

 

Try This https://plugins4free.com/plugin/849/

 

Last edited by a moderator: Jul 26, 2022

I do 1dB or less, per instance. Preferably several doing little than one doing a lot.

I learned to clip the converters in 1999 by a mastering engineer. I was like "What?! Are you clipping/distorting the input of a DAT machine?! That is wrong! It will distort!". He just laughed and said "not if you do it right and in stages". He then showed me how he clipped and limited at the start of the chain and then doing the same at the end.

There's also the technique of using transient designers to bring back transients that have been clipped/limited (before going into the final limiter/clipper). Lol.

 

Last edited: Jul 27, 2022