Improving gross overload

[Ron Cooper]

I have been given a grossly overloaded file and while I realise there is not a lot you can do to rectify this, I was wondering if there just happened to be some software now that that may give me a chance to even slightly improve it over the usual audio restoration tools. I have Sound Forge 9.

RonC.

[Paul DeBaets]

Ron, I have had great success with Isotope 3. You can try out the software for 10 days free.
iZotope RX 3 | OVERVIEW

Paul

[Rick Reineke]

If you can post a clip, we could better render an opinion. Digital overs are one thing, preamp overdrive is another. If it's the latter, nothing's going make it pristine, not even the advanced version of iZotope RX3.
Sound Forge Pro 11 does come bungled with a iZotope restore that works fairly well on analog distortion.
Check out the SCS SF webinars.

[Zoran Vincic]

I work in audio production too and I have some tools for audio restoration at disposal in my studio.

PM me or post a short sample here (10 seconds will be enough, no mp3, uncompressed wav please) and I'll PM you back after I give it a try.

[Ron Cooper]

Thanks all. - This is a wonderful forum. Hope this attatchment works ! - If it does not, can you please tell me how to do it.

RonC.

[Don Bloom]

Aww, never mind...Sorry about that.
Was going to respond and just had a huge brain freeze. :-(

[Rick Reineke]

Don't see it Ron.
At the bottom of the reply window select "Manage Attachments"
Select the desired file from your computer and click "Upload". If that doesn't work use Dropbox, 4-Shared, YouSendIt, Google Drive or other file sharing site.
I have previously not attached audio files, the attached is for confirmation.

[Ron Cooper]

Ah, thanks, now I see how to do it. - I was looking in the wrong spot.
So hopefully here's the upload.

RonC.

[Zoran Vincic]

I'll give it a try later, but I can tell you right away that it will be full of artifacts as it's so distorted.

[Greg Miller]

Nice clean recording of the room tone...

[Battle Vaughan]

OOOh,my goodness, this is a tough one! Lots of artifacts, to be sure. I'll offer what I could do in the hope that somebody does lots better and I can learn from them:

My tools:
Adobe Audition CS6. Heavy Declip function using Fourier 128bit transform. Followed by level adjustment, then noise reduction to remove the loud room tone, followed by EQ, rolling off below 200hz for boominess and gently rolling off above 1500 hz trying to remove some of the sibilance. Then fairly heavy compression to try to level out the amplitude.

It's some better, but that's a real damaged track, I'm afraid. Comments?

[Greg Miller]

Mr. Vaughan,

That's better than I would have thought possible. But, out of curiosity, I would very much like to hear it without the denoise function.

Thanks.

[Battle Vaughan]

I'll go back and try without the denoise, and post it, but I can tell you the room tone is fairly overwhelming, and drowns out the voice portion.. Taking out that random noise made the voice much more audible and the breaks between words more clear. Without doubt it had a significant effect on the voice frequencies, leading to the somewhat robot-voice effect we are left with...but I'll pop up a sample soon as time permits...thanks for your feedback!

[Greg Miller]

Thanks,

The robotic nature of the voice is what I'm curious about. I was wondering whether the track would be more intelligible without the heavy NR.

I tried several variations of LPF before and after de-clipping, but never got a result that sounds even vaguely like a finished track. The resulting track might be useful for manual transcription, if you want a written record of what was said, but certainly not even close to good audio quality.

Interestingly, while my de-clipped version looks quite different from the original, there is often no significant difference in intelligibility! (This makes me wonder whether the human psycho-auditory mechanism has a de-clipping function.)

[Battle Vaughan]

Here are three versions: straight heavy declip using Fourier 128 bit; same with denoise added; same as #2 with eq added to rolloff below 200hz and above 3000hz (trying to get rid of the 'spatter' artifacts with limited success). This is a toughie. But you're right, I think the heavy nr did more damage than good, this voice is less robotic. This time around, the room tone was less intrusive, I think I may have amplified it with something I did in the first version, about the level-adjust stage. It was like a freight train in the room before n/r in the first version.