frequency cut/remove/? noise.

Have problems with some of my CEP original recordings. Always left channel, never on audio played by other audio sources--CD's, movies, WMA, RA, etc. files. Just CEP originals. wav’s.

Just digging into frequency clean-up, so not well-informed.

I've found that the sound of the "noise" is due to frequency issues--an acoustic piano is come cases. I'll use 1 of the 2 situations where I have an acoustic piano, sometimes alone sometimes with acoustic guitar. I'll use the latter as my example here.

Now playing around with frequency to handle the problem, I've isolated the freq range to 584-595, narrowing that range produces sub-optimal results.

Doing a cutout on scientific filers for that range cuts the problem out, sonically. Leaves a click or two that can be dealt with. Fairly slight attenuation to the whole section overall, but not very much. Different results using different filter selections

Using notch filter, running it twice with the first 6 freq. and then the second 6 it doesn't get enough at attenuation of 5db, but works at 10. Oddly it attenuates most of the selection, but actually boosts some of it (I’ve noticed this with some other "attenuation" effects, like NR). The other 6 attenuates much more. Couple of clicks, again.

I try it in parametric EQ 2 diff ways: 1 way has been to hit the bookends--584 and 595--with minimal attenuation of approx -1. That works, but produces relatively large overall attenuation--far more than other ways I can get at that with master boost, but not sure about the wisdom of doing that. The other way I've tried is using 589 or 590 and then putting the width at 5--wasn't up to speed on Q width before. Works, but with large overall attenuation of resulting wave--roughly need to boost entire wave selection by about 2 db to get back to normal-ish.

Questions, keeping in mind--this is part of a very informal set for a CD for no commercial purpose and just to be given to a very few friends, i.e., not much at stake here:

A.  Is reducing level of a frequency(ies) like this a good way to go? (Really don't want to re-record--the piano is on the verse progression once in the song w/out vocals.)

B. If so, is reducing a small range like I'm trying a good strategy?

C. What are the pros and cons of each method? Is there one that is most appropriate, or is it matter of just listening to end results? I’m not so sophist iced, so there likely are better ways to handle this and within various methods.

D. If I am going to do it, shouldn’t I apply it to the whole track? And at that, should I also apply it to the right channel track as well, even though there's no problem with it?

E. I'm assuming that I should attenuate only as much as keeps the sound from being a problem, but as I've already finished the song months ago with applying hard limiting as my last step, wouldn't there be a problem with reducing too much (which I'll need to do to remove the problem) that I'd need to boost the volume enough to match the other songs on the rest of the CD?

F. Applying the filter(s), reduce the relevant selection by much more than the overall track, so bring the entire track up to relative par db-wise would mean applying different levels of boost for different parts. Seems potentially problematic for overall sound, depending on how to boost the whole song; i.e. HL could even things out better if things boost enough to hit the max. often enough.

F. I’m having some noise of slight crackling for some other songs—left channel only. NR, click, isn’t working (incl. Sonic Foundry). Wondering if there might be certain frequencies where high crackle is most common that I could attenuate slightly—just as hiss is found in particular places?

Lots of questions, but what I learn about this stuff will help me greatly in cleaning up other songs and esp. handling frequency issues in general.

All help greatly appreciated

Using notch filter, running it twice with the first 6 freq. and then the second 6 it doesn't get enough at attenuation of 5db, but works at 10. Oddly it attenuates most of the selection, but actually boosts some of it (I’ve noticed this with some other "attenuation" effects, like NR). The other 6 attenuates much more. Couple of clicks, again.

Sometimes if you have audio in the filter skirt (the sides of it) and use a high-order filter like the notch ones, then yes you will get a boost of frequencies at the sides of the ones you are trying to filter out - this is quite normal, and happens with 'real' filters too, often more so because the passband ripple is worse.

If you can find a specific frequency, or small range of them, and reducing them improves the sound, that's fine. I wouldn't use the notch filter for this - generally experimenting with the parametric is the best way. The easy way to spot the frequency you are looking for is initially to set up a narrow boost (10-15dB with a narrow (high) Q setting), and sweep it across the band until you hear your problem area get a lot louder - this is much easier to spot. Then alter the boost to a cut, and experiment with the Q and depth settings. In general go for the minimum cut you can get away with. If you can't hear a difference, then just do it to the channel that's affected. If it's caused by some sort of resonance in the room when you recorded, then you might want to reduce the whole file with the settings you find that work. But if it's only a few notes, then I'd do it by hand.

If you have 1.5 then the marquee selection tool in spectral view is another good way of getting to these issues and solving them quickly and easily - quite often the specific resonances show very clearly this way, especially when you zoom in on the affected area. With any sort of signal that's anywhere near close to full modulation, then you don't really want to boost everything except the problem area - you stand a good chance of getting into an overload situation this way - cutting is the norm here.

All crackling is different, and what fixes it rather depends... have you tried using the declicker that you'd normally use for cleaning up vinyl transfers?

Most of these small niggling problems are best fixed at source, ie when you make a recording in the first place. It's always worth paying quite a bit of attention to monitoring what you are actually getting when you place a mic - locating the source of rattles, squeaks etc at that point is always the most successful way of dealing with them. Time spent at this stage invariably pays dividends later.

Steve, thanks much. My omission to have left out that I'm using CEP 2.0a.

For parametric, what are suggesting for denoting specif frequencies? A high and a low to get at the range, or multiple points in between?

The crackle problems and and the high freq issue affects only a few tracks out of 19. This very moment it's occurring to me that the 2 main songs I can think of weren't recorded on CEP. 1 started on a 4-track, and later imported to a cheap $30 audio program that I started out with in late 90s. One of the others was completely done on that program and then dumped to tape. Then imported those songs into CEP. There are some parts of other songs done entirely in CEP.

All of the problems are left channel only. Considering original sound sources for at least 2 songs, but some from CEP, it makes me wonder if there's something wrong with my soundcard that would have affected these few recorded in CEP songs and then the others based on applying effects post-hoc. The left channel, different original sound sources makes me wonder if the problems relate to something, some cases at least, other than merely recording.

By handling things by hand, I take that to mean not applying to a whole track or section, nor to the corresponding places on the other channel. Correct?

What applying a little bit of reverb to the crackled songs to mask the crackle a bit, smooth it over?

Steve, thanks much. My omission to have left out that I'm using CEP 2.0a.

Unfortunately for you, that rules out the Marquee tool...

For parametric, what are suggesting for denoting specif frequencies? A high and a low to get at the range, or multiple points in between?

Ideally, if you can do it in one instance, that's better than multiple points of attack - these generally cause more anomalies if you get them too close. If you have problems that are well-separated then obviously you can go for more than one. The thing to avoid generally is the skirts 'meeting' unless the intersection slope is very gentle. In general, general slopes cause less response problems anyway - it's the incisive spot attacks that are likely to be more noticeable if there's a lot of sound in them.

The crackle problems and and the high freq issue affects only a few tracks out of 19. This very moment it's occurring to me that the 2 main songs I can think of weren't recorded on CEP. 1 started on a 4-track, and later imported to a cheap $30 audio program that I started out with in late 90s. One of the others was completely done on that program and then dumped to tape. Then imported those songs into CEP. There are some parts of other songs done entirely in CEP.

All of the problems are left channel only. Considering original sound sources for at least 2 songs, but some from CEP, it makes me wonder if there's something wrong with my soundcard that would have affected these few recorded in CEP songs and then the others based on applying effects post-hoc. The left channel, different original sound sources makes me wonder if the problems relate to something, some cases at least, other than merely recording.

Can't really comment on this without hearing an example - there really are a lot of different types of 'crackle' - some are relatively easy to recognise and sometimes fix, others aren't. But anything that's been through multiple processes, especially if it's come from a 4-track, has to be a potential victim for anything in the path, I'd guess.

By handling things by hand, I take that to mean not applying to a whole track or section, nor to the corresponding places on the other channel. Correct?

Yup...

What applying a little bit of reverb to the crackled songs to mask the crackle a bit, smooth it over?

You could try this, but don't be too surprised if you end up with reverberated crackle too. In other words, you could easily make things worse rather than better. If you can post a short example of the crackled track, it would be much easier to make sensible suggestions about dealing with it, because then I (or anybody else who fancies it) can try a few things, and see what works best. If you do this, just make it a small section, and ideally make it a wav file rather than any compressed format.