iZotope RX has much better clipping restoration than the Sony Noise Reduction, but the Sony product does work. I have Sound Forge but haven't upgraded in a long time. I think this may now be included in SF, but someone else will have to correct me.

Try this:
http://www.cutestudio.net/data/products/audio/seedeclip/index.php

Sound Forge does now include Clipped Peak Restoration, and it does work. Sometimes.
Its definition of clipping is pretty much restricted to truly flat tops at 0dB.

"Its definition of clipping is pretty much restricted to truly flat tops at 0dB. "

True indeed however this is easily fixed. Apply just enough gain to get hard digital clipping and then use the Clipped Peak Restoration FX. I've also used this technique to get useful repairing of analogue overloading.

However do not expect miracles. If you've seriously clipped the audio that's it, it's fried, toasted and burnt on both sides. Also the degree of repair possible depends on the nature of audio. Complex sounds will loose a lot of the original data when clipped.

Bob.

True indeed however this is easily fixed.

That's true, too, if we can all agree to be loose with the definition of "easily." Heh.
I have successfully used that technique at times, but sometimes it requires a painstaking surgical approach.

Another Sony limitation is that it makes all the clipped peaks perfectly round.
OTOH, RX guesses at what the peaks would look like in real life. Whether or not the results are really as they would have been without the clipping, I have no idea. But they look real. Heh again.

Of course, the answer is to not clip on recording.

One technique I use when the audio feed is mono, and I have two channels available, is to set the gain on the second channel to 9 or 12 db below the first.and feed the incoming to both That gives me the additional headroom just in case it's needed.

Of course, the answer is to not clip on recording.

I just restored a tape recorded in 1950, which had amazing amounts of hum and clipping. I tried to find the recording engineer so I could fire him ...

That wasn't me! I swear... although, does anyone here remember paper backed magnetic recording tape? unless you set the level for clipping on those, the signal would get buried in the background noise. Those were the days, louder was always better, even the marketing hype was there: "Our amplifier's volume controls go to 11, where all others go only up to 10."

That must have been some recording engineer. According to my history books digital recordings didn't exist until decades later so clipping wasn't possible in 1950.
What I do find amusing is the amount of money some will spend to recreate the sound from back then of magnetic devices pushed beyond the limits of the B/H curve.

Bob.

According to my history books digital recordings didn't exist until decades later so clipping wasn't possible in 1950.Believe me, it is very easy to saturate almost any analog amplifier design, whether tube base, bipolar transistor, MOSFET, class A, B, C, etc. You can feed it "11," but if it's only got "10" in the tank, that's all you're going to get, and the result is a clipped peak. What's worse, you usually also get distortion as the amp nears saturation and, unlike peak clipping, you end up with intermodulation distortion which is ugly to listen to and impossible to remove.