Take the 2-minute tour ×
Sound Design Stack Exchange is a question and answer site for sound engineers, producers, editors, and enthusiasts. It's 100% free, no registration required.

I have some old voice recordings I would like to restore and I'm not sure what software I should use and in what order I should apply the different transformations.

They were digitized a while back and I only have access to the mp3 version. Some of the files are louder than others (and noisier), The noise is mostly hiss, not clicks or pops, and the voice is kind of mumbled, which I'm guessing is because of the very lossy mp3 format used.

What I'm trying to achieve is less noise, uniform sound level and, if possible, clearer voice.

My guess is that I should use a normalizer or compression to smooth the audio level and apply a noise reduction filter but I'm not sure if the noise reduction filter works better before or after the normalization step.

Also, this may sound dumb but I was wondering if it is possible to "enhance" the consonants so that the voice is easier to understand.

If any of you have ever restored this kind of audio, I'd love to hear about the process.

P.S. If a moderator should see this question, it might need some better tags, like noise-reduction or audio-restoration.

share|improve this question
add comment

migrated from avp.stackexchange.com Jan 24 at 12:01

This question came from our site for engineers, producers, editors, and enthusiasts spanning the fields of video, and media creation.

3 Answers

You can do a lot of this "by hand", such as it is, by using a graphic EQ and simple amplification matching.

For enhancing the consonants so the voice is easier to understand: the most critical region for speech comprehension is in the 2 to 4 kHz range. Boosting in that range will likely help.

That brings me to my next point as far as noise reduction: if you have broadband, overall noise, I would recommended inserting a graphic EQ and reducing every band that's NOT between 1 and 4 kHz as a starting point. It will probably sound like a telephone conversation, but in there you have the information you need. You can gradually add back signal content moving out from the center, and pay close attention to when you've "added" back an unacceptable amount of noise. You can play with it from there.

If you have more tonal noise (fans, etc) you can use a parametric or graphic EQ for filtering.

General noise reduction algorithms and plugins can work, but they can be expensive and leave undesirable artifacts. Give it a shot! You might be surprised as to what you can get with just a graphic EQ.

share|improve this answer
add comment

Consider trying:

Both of these are billed as audio restoration/noise reduction programs, and serve the kind of purpose you're describing.

share|improve this answer
    
But both are specialised on analog artifacts and won't work ideally if the quality is further degraded by bad mp3 compression. –  leftaroundabout Jun 9 '13 at 16:37
add comment

Not a fan of waves. iZotope is strong. Sonnox/Sony Oxford/Sony Sound Forge is best in breed for my money.

The feature you're looking for is adaptive/fft noise reduction - you should see somewhere a "learn" button. This means that you find a section of the recording that is "pure noise" and play it while the program automatically analyzes the frequencies. Then you "apply" the "noise profile" to the entire recording and the noise is (hopefully) gone.

share|improve this answer
add comment

Your Answer

 
discard

By posting your answer, you agree to the privacy policy and terms of service.