Recorded a ground I can't get rid of

[Rokket]

Unfortunately I think it's difficult to make any specific diagnosis from that screen shot because I can't see the frequency scale which I assume is along the bottom of the graph. Without being able to verify the actual frequencies of the various peaks, it's not a complete help.

Is that what it looks like when there's nothing but hum? Because that's what we really need to see - along with the frequency scale. It looks like *maybe* there's at least some regular repetition there, but I can't say for sure yet.

G.

Leet me doctor up the pic and repost it. And yes, this is during a section where it's only the hum. She paused for dramatic effect or something like that...

 

[SouthSIDE Glen]

Leet me doctor up the pic and repost it. And yes, this is during a section where it's only the hum. She paused for dramatic effect or something like that...

OK, cool; it's past my bedtime here and I'm falling asleep at the keyboard, so I'll check out the modified screen shot first thing in the morning...maybe Gecko or someone on the other side of the world where it's probably daytime might chime in in the meantime.

G.

 

[Rokket]

Is this any better?

 

[bennychico11]

I'll see what I can do...

i meant an audio sample...do you have a portion of just the hum...or even just a 10 second clip with hum/noise and talking to see if a NR plugin is even worth the effort?

PM me if you need an e-mail address to send it to.

 

[SouthSIDE Glen]

Is this any better?

OK, now I can read the numbers...but only after dragging it into Photoshop and adjusting the levels. Man, I've been discovering lately that I seem to keep my monitor much darker than most other folks; I don't see how you guys can stare at such bright lightbulbs so long .

I assume the yellow graph represents the hum. I'm not sure what that magenta line is all about. Anyway, Rokket Man, assuming the yellow line, that's quite a bit of noise you have there, and it's not just a simple fundamental and harmonics. You do have a small series of fairly wide bumps at approximately 150, 300 and 600 Hz which could be related, with another possibly related one at about 750Hz, though that's kind of a stretch. Everything else above that seems fairly noisy to me, with no obvious patterns sticking out.

The fact that the noise runs all the way from DC to approx, 7-8kHz, with not much pattern above the upper bass is bothersome to me in that a NR plug like what Benny wants to try may take out a lot of the signal as well. But the only way to know for sure is to give it a shot.

Other than Ben's approach, which I would still pursue, what I'd probably do in my circumstance would be to throw RND's Uniqualizer at it, allowing me to run several simultaneous EQs and filters at it in one pass. I have one run a high pass at about 150Hz or so, and then set up a harmonic filter with fairly wide Q to match the bump widths in the graph with a fundamental around 150Hz (adjusting the exact value to exactly fit the curves).

Then with what's left I'd either run it through NR* or I'd hit it with an expander set to maybe about 1:3 and a threshold of about -70dB.

*Note to Benny: I have had situations in the past where, if the NR winds up artifacting the good signal too much, that sometimes just taking out some of the obvious stuff with targeted EQ first is good enough to let the NR take care of the rest w/o artifacting the signal. Of course, if you're not getting artifacting, then don't worry about it .

G.

 

[bennychico11]

*Note to Benny: I have had situations in the past where, if the NR winds up artifacting the good signal too much, that sometimes just taking out some of the obvious stuff with targeted EQ first is good enough to let the NR take care of the rest w/o artifacting the signal. Of course, if you're not getting artifacting, then don't worry about it .

G.

exactly what I usually do. No point in adding a resource hog that's just going to add unwanted artifacts if I can reduce as much as possible to start with an EQ plugin. Typically I notch/filter out everything I don't need (highs/lows/hum/whines) then run the rest through one of the NR plugs. It's never perfect, but a lot of times is better than nothing. Of course, if the noise is louder than the talking...I can already guess not much will be done.

btw, hope things are well with you Glen.

 

[SouthSIDE Glen]

exactly what I usually do. No point in adding a resource hog that's just going to add unwanted artifacts if I can reduce as much as possible to start with an EQ plugin.

I had a situation that I talked about here before quite a while back where I had a guy who was doing some synth-based work (similar in style to Jean Michael Jarre) and was running through a noisy sound card. It sounded like your typical pinkish noise but when I tried to remove it plain and simple with fingerprint-style NR (I tried two different ones), it artifacted the signal so badly that it turned a synth string section into a sick whale calling out for it's mother, and that's not really an exaggeration. It was really quite bizarre.

An FET analysis showed a very strange harmonic series with a fundamental around 384Hz or something weird like that embedded in the noise that climbed all the way up the scale. When I attacked that series with a harmonic filter first, the NR worked fine on the rest of the noise and barely touched the musical signal at all.

I still have no explanation as to why his soundcard generated that weird harmonic, or even if that's where it was actually coming from (it wasn't from his synths, we checked that), nor why that would cause the NR's such incredible trouble; but there we have it.

btw, hope things are well with you Glen.

Same boring ol' story around here. It's good to see you back posting once in a while again, my friend. I hope this doesn't indicate trouble with the situation that put you on self-imposed exile to begin with.

G.

 

[bennychico11]

I still have no explanation as to why his soundcard generated that weird harmonic, or even if that's where it was actually coming from (it wasn't from his synths, we checked that), nor why that would cause the NR's such incredible trouble; but there we have it.

What was the soundcard? Onboard?
I did a test that I posted around here awhile back of sending a 17kHz tone out my stock card and into my PT system and just the playback introduced so much extraneous frequencies.

Same boring ol' story around here. It's good to see you back posting once in a while again, my friend. I hope this doesn't indicate trouble with the situation that put you on self-imposed exile to begin with.

G.

just trying to keep busy. Things were slow for us for awhile due to the economy.
I just lost interest in the forum world for awhile and have only recently started to come back. But I still tend to hang out on the PT forums for the most part...and a few other non-audio ones. That, and the girlfriend takes up most of my time when I'm not at work

 

[SouthSIDE Glen]

What was the soundcard? Onboard?
I did a test that I posted around here awhile back of sending a 17kHz tone out my stock card and into my PT system and just the playback introduced so much extraneous frequencies.

Yeah, it was whatever SoundBlaster they included from the factory. Shortly after that incident he replaced it with an eMu 404 (I think that was the model number). What a difference for the better that made. Sure, it's no high end interface by any long shot, but the difference between the two was distinctly audible.

G.

 

[Rokket]

I'm not sure what that magenta line is all about.

That magenta line showed up when I clicked on the "show phase" tab on the analyzer.

I have a plugin that will allow me to run several filters across the track at once. I'll give your recommendations a shot and see what happens. Messing around with it yesterday before you posted this, I had already tried that plugin out a bit. I got it close to being gone, but I lost a lot of volume on the track. When I boosted the dB level, all the noise came back with the vocal.

 

[SouthSIDE Glen]

That magenta line showed up when I clicked on the "show phase" tab on the analyzer.

Aaaah, OK. Very slick. Not relevant to our situation, but a very nice thing to have in the tool box.

but I lost a lot of volume on the track. When I boosted the dB level, all the noise came back with the vocal.

It's tough because your noise is rather broadband in nature, with the peaks themselves being fairly fat as well. Scoop out too much, and you're talking away too much of the signal as well.

I suspect your situation may wind up being a compromise in the middle somewhere, removing some of the worst of the hum, but leaving some in there in order to save the signal, then hope a good fingerprinting N/R plug can handle the rest OK.

Here's the trick, and there is no one answer, you'll have to play around a bit: finding the optimal combination of bandwidth and amount of cut that will give the best compromise. Bandwidth meaning how wide each of your EQ cuts at each of the frequencies is - too wide and you might scoop out too much signal, too narrow and you might not take enough of the noise. And amount of cut referrs to to the depth of the cut itself with, again, finding the right balance between noise and signal.

*In general* (with exceptions) the wider the cut, the shallower the cut, and the narrower the deeper (though it doesn't always have to be deep.)

G.

 

[Rokket]

Aaaah, OK. Very slick. Not relevant to our situation, but a very nice thing to have in the tool box.It's tough because your noise is rather broadband in nature, with the peaks themselves being fairly fat as well. Scoop out too much, and you're talking away too much of the signal as well.

I suspect your situation may wind up being a compromise in the middle somewhere, removing some of the worst of the hum, but leaving some in there in order to save the signal, then hope a good fingerprinting N/R plug can handle the rest OK.

Here's the trick, and there is no one answer, you'll have to play around a bit: finding the optimal combination of bandwidth and amount of cut that will give the best compromise. Bandwidth meaning how wide each of your EQ cuts at each of the frequencies is - too wide and you might scoop out too much signal, too narrow and you might not take enough of the noise. And amount of cut referrs to to the depth of the cut itself with, again, finding the right balance between noise and signal.

*In general* (with exceptions) the wider the cut, the shallower the cut, and the narrower the deeper (though it doesn't always have to be deep.)

G.

OK! I've used up enough of your time on this. I think you have given me enough to go on. If nothing else, when she does do something with this presentation in a month or so, she will appreciate that I tried to do a good job in the very limiting parameters she set for me...