That got me thinking about an idea I had long ago..
A setting to specify and maintain any desired headroom for automatic and transparent 'over' protection without using a traditional limiter or by writing two redundant files recorded with different gains-
Instead, I propose a feature that automatically lowers gain by a set, user-defined amount anytime the input level crosses a user-defined threshold and notes each occurrence in a simple CUE file with the same name as the WAV file. After recording, the CUE file could then be used to specify inverse gain changes at the appropriate points. The gain reduction applied could either be held until a new file is started (an effectively infinite 'release time' in compressor/limiter terms) or in a more advanced implementation, a release time could be specified and that event also noted in the CUE file.
Example: Say I set levels so that the program peaks I expect reach -10dBfs. Say I also turn on the over protection feature and set the threshold described above to -6dBfs, specify a gain reduction of -3dB for any peak event that crosses that threshold, and specify a release time of 'hold until start of next file or manual reset'. The recorder then acts normally for all expected program peaks of less than -10dBfs and also for any program peaks of up to -6dbfs. But if an unexpected event peaks at -2dBfs, over protection automatically reduces my gain setting by -3dB for the remainder of the recording and notes the exact time this occurred in the CUE file. Effective headroom is then increased to 13dB for any following events. Any following peaks higher than -6dB will again kick in over protection and further reduce gain by 3dB increments, each occurrence noted in the CUE file. The gain will rapidly be adjusted to a setting better suited the program material and the end result after re-adjusting the levels of the file afterwards using the CUE file entries will be completely free of any traditional limiter or compression effects.
A really large initial peak of more than 10dB (or subsequent peaks of increasingly higher +3db increments) will still cause a single, brief over, but over protection immediately kicks in and reduces gain each time, so that the gain is rapidly and automatically adjusted downwards to accommodate the unexpectedly high signal levels.
Such a feature would make balancing noise-floor against peak level simpler by providing peak protection without sonic drawbacks. It could be invaluable for recording very low level or highly dynamic material the user is unfamiliar with. I’ve recorded things with overly-conservative levels just to completely avoid the chance of distortion in unfamiliar circumstances and ended up with files that peak at -25dBfs and RMS values of much less. This feature would have no sonic ill effects if it was left ‘on’ for every recording, regardless of if it was actually needed or not.
In addition, the CUE file could also easily note other events, such as user specified event markers, low battery voltage warnings, etc.