I have an audio recording of people whispering that would llikely be audible were it not for the white noise on the recording. I have heard that people can do remarkable things with sound editing programs. What are some programs that might be able to help me hear the words being said on this recording? What are some good programs (free if possible) for cleaning up white noise? Also, if anyone is really handy, would they be willing to do this project for me? Thanks in advance! Super Mario Bros. - The Wicked Star Story "Ah, the proverbial sad sack with a wasted wish."
Originally posted by: iRantanplan Isn't white noise supposed to be inaudible?
I think I know what you're thinking of, but I don't remember what it's called. Probably another word for ultrasonic or infrasonic.
Originally posted by: andy_k_250 I have an audio recording of people whispering that would llikely be audible were it not for the white noise on the recording.
I haven't spent a lot of time on audio, more time reading about it. So I hope an expert steps in, here.
The signal, probably, barely exists - near the level of the random tape/circuitry hiss. Worth a shot, though.
Goldwave (fully-functional nagware) has noise reduction. But the more heavily you apply it, the more metallic and robotic the sound gets (like a special effect).
Sony (noise reduction plugin). Sony bought-up Soundforge, which was the big name in audio software. Dunno how its doing under new ownership. Look around their site. Dunno if their trialware is fully functional. You might be able to do a temporary licence transfer with a pal...
If no one, more knowlegeble than I, answers you, here, you'll need to google around, try stuff, and look for forums, I guess.
The first thing is to get a noise profile. Circuitry hiss isn't entirely random, and there's likely ambient (background) noises going on. So you take a stretch of audio where no-one's talking, and have the software analyze it.
Then there's tricks with using equalizers, bandpass filters, and bandnotch. The strongest harmonics in human voices are in certain bandwidths. CB's, and such, filter to a narrow bandwidth for minimum "recognizability" (but you know how badly distorted those are). I haven't had the slightest luck with that. There's vocal on every bandwidth. But if you do filter to the bands where the vocal is strongest, you eliminate a ton of noise.
If it's stereo, and the people aren't directly between the microphones... I've had a crazy idea for a long time...
I wonder if you could experiment with delaying one channel (microseconds), so that the voices would land exactly on the center channel? I haven't ever got around to to trying it, but it should work... Then the two channels would reinforce each other, while the static would be spread around in all directions.
Another crazy idea would be to try a "tube warmer" plugin, and set it to apply to low volume, too. I'm talking about plugins that emulate the harmonics effects of old Tube-Driven amplifiers, when the amps were driven high. That adds pleasant, and natural-sounding harmonics to the sounds, and I expect it would make voices more audible. And, while the static would get more harmonics, I suspect the "amplified" static wouldn't obsure the voices any worse. Back in the day, Tube amps made things sound a lot sweeter and clearer, when you cranked them up. Anyway, one thing Tube amps excelled at was making limited-bandwidth sources sound more natural, by adding those pleasant harmonics. So if you got a good result by harsh equalization, you might make it a lot clearer with the right "tube" settings.
If you have the original tape, then the original tape recorder would be set up to play it back the best (unless the manufacturer cheaped out too much, which is common). It'd be great to have a pro tape decks that let you adjust the azimuth, bias, and all that good stuff... I could use one of those, but I've been afraid to look at what one would cost (even used)...
I noise reduce my films a lot. I use Noise Reduction in Audacity (Mac) or Cooledit (Win), at the lowest setting, and create a new noise-reduced WAV. Sometimes I noise reduce twice and save it again also. The result tends to sound very electronic. However it will sound better when you mix your one or two noise reduced tracks together (in Premiere or FCP) with the original not noise-reduced WAV (added in at a lower volume of your choosing). Also, on the noise reduced version you can lower the volume to nothing on very quiet parts of the audio, which tend to sound more electronic than when people are talking loudly.
The results of this in my movie Gods of Los Angeles were quite good overall.