extremely low sound-level need to boost

[jackstpaulUHS]

I use CEP 2.1a

A problem I have now, and have had before on other stuff, is when there are small sextons that are very, very low signal-level, almost inaudible. The issue I have is trying to boost that portion's level without major noise problems, etc.

What I'm working with is a one-channel track from a 4-track. I was recording an acoustic guitar track thru a regular microphone, and preceding it were some comments between myself and my friend who was far away with the 4-track itself. Some of his talking was picked up at a high-enough level that I can boost them without a problem. But then there's a few second section where it's barely audible--he must have walked further away or turned away so that the mic didn't pick it up well. My talking was picked u at a much higher level being right in front of the mic.

Boosting the barely audible part results in a very significant amount of noise that makes it not worth using--just for that section.

I use sonic-foundry's NR plug-in and CEP's noise reduction function, and no matter what I do to clean it, the resulting sound is really, really distorted from the sound from the rest of it, though cleaned greatly of the noise. I assume I’m losing some frequencies (levels) to result in the distorted sound.

To use that part in that condition makes for a really awkward change in sound that's too off-putting.

I've done a lot of stuff like this in cleaning boombox recordings, but this one has me stumped.

The sound level is so low before boosting that NR isn't possible. I've boosted by regular amplification an, HL, and various dynamics processing--compression, limiting.

As it's been a problem before, I’m interested in hearing comments about this and the issue in general. Is there anything in general that can be done other than what I've tried? My hunch is that it's a lost cause.

Any explanations and/or suggestions?

 

[suprstar]

If your track is so quiet that it's right at (or barely above) the noise floor, youre screwed.. You'll have to re-track it. Compressing will just make the noise louder, and there'll be no peaks for a limiter to catch.

 

[jackstpaulUHS]

Yep, I’m screwed. Can't re-do. It was a brief snip of dialogue with someone now departed.

My thought was always that if the sound is louder than the noise that you could clean enough of it away within reason to make it workable, but this situation is beyond repair.

Can you just expand (no pun intended) on your comment about peaks and a limiter? I think I know what you mean, but fleshing it out a tiny bit might help my general knowledge. The only use I make of a limiter is hard limiting a track/song to boost the signal to max desired level and level off the peaks relative to whatever degree I want—the last step in my process.

Thanks

 

[lpdeluxe]

It takes a lot of experimenting with different strategies and a lot of patience, and that might not be enough

A couple of years ago a friend of my wife asked me to transfer a home recording to CD for her. The home recording was made in April, 1946 onto a blank lacquer disc via a cutter: no wonder tape shortly thereafter took over the world!

The resulting vinyl record, which consisted of the lady (then 5 years old) singing "Shortnin' Bread" twice into the mic and engaging in repartee with an unidentified grown woman, had not been stored properly. It sat in its paper sleeve in various boxes in various attics until its owner found a victim (me) to transfer it. It was full of scratches and general wear and tear.

It was recorded at 78 RPM. I didn't have a turntable that had that speed on it, so I played it back at 45 RPM through a phono preamp into my digital mixer and Adobe Audition. After recording, I used the "stretch" function in AA (also in CEP) to brig it up to pitch. I couldn't do anything about the RIAA curve differences, but at this point I was pretty happy just to have a digital file of the record.

Then came several evenings of trying Click and Pop Reduction, Noise Reduction, Scientific Filter (used as a highpass to eliminate rumble) and on and on until the singing was intelligible. I never could lift the adult's conversation (off-mic) out of the noise, so I'll never know what she was saying. Luckily the lady was satisfied with hearing her own voice and left a happy client.

Good luck.

 

[mixsit]

..Compressing will just make the noise louder, and there'll be no peaks for a limiter to catch.

Basically he's saying reducing the dynamic range isn't the direction to go. The thing about the limiter could be skipped all together.

Since you're in the computer, how about zooming in with volume automation? It'd be like gating/expansion, but where an expander would need to see actual volume differences and there are none (due to the low signal to noise) you would be picking and tailoring the highlight spots?