Reversing effects of over-attenuation/limiting

[Ian Stark]

I have some interview audio which has been recorded way too hot, either at the camera or at the wireless end, not sure which (not my audio). The waveform is basically flat throughout and it sounds ghastly.

I'm not expecting miracles, but is there anything that can be done to partially rescue such audio? Would an expander help?

All advice welcomed.

[Steve House]

Expander not likely to help. The problem isn't that the difference between the lowest and the highest level signals is too little, it's that the tops of the waveforms have been sliced off, turning them into dirty square waves that generate oodles of hashy-sounding high order harmonics when played back. You need to lower the overall level and restore the waveform tops. There is software available to do that but how successful it is depends on how severe the problem is to begin with - there's only so far that software can go towards accurately guessing what information was lost in the clipped off portions. Take a look at SoundForge and Audition, they both have clip restoration tools. Another good one is Izotope's Rx. Search Google for Declipper and Clip Restoration.

[Ian Stark]

Steve, thanks for such a helpful response. I have an older version of Sound Forge (I mean REALLY old). I'll see if I can upgrade. I did look at Rx some time ago, specifically to remove aircon hum etc. I ended up using Waves X-Noise which has proved to be excellent (but not for this job, obviously). Time to look at Rx again I think. Seems to be very highly regarded. Thanks again.

[Greg Miller]

Ian,

When you say "the waveform is basically flat," is that just when you're zoomed out and looking at several seconds of the file?

The important question is: how does the waveform look when you zoom in enough to see individual cycles of the sound? (I'm thinking that the entire horizontal axis of the display would be around 10 msec to see this much detail.)

If you look at the individual cycles of the audio, and the tops and bottoms are clipped perfectly flat at 0dBFS, then you've got digital clipping in the recorder. Audition can repair that, but only to a limited extent.

If the individual cycles are somewhat flattened, but at a level lower than 0dBFS, then the clipping probably occured earlier in the chain, and the audio might have other problems more complex than the flattened waveform. In that case, all bets are off.

You might want to submit a 30-second sample of your file so we can look and listen.

[Ian Stark]

Thanks Greg,

I'm not in front of it right now but as soon as I get a chance I'll check that out and respond. Thanks for the comment.

I completely realise that miracles aren't going to happen here, and I'm fairly sure that the clipping occurred at the radio mic, based on what the girl who shot the piece told me, so I fear the worst.

Anyway, I'll come back with more info as soon as I can.

Cheers.

[Rick Reineke]

As Greg stated, if it's a question of moderate digital 'overs', clipped peak restoration software can often help, depending on the severity. If the problem is an over driven pre-amp, that's usually not salvageable, at least to the point of making it sound decent.