Q Is it a problem using commercial tracks as a reference if audio files are clipped?

Sound Advice : Recording



Hugh Robjohns


I’ve put together a group of commercial reference mixes. I chose them because I liked the way they sounded, but I noticed that two of them had been clipped. Are these mixes still worth using as references? And how can they clip without distorting in an ugly fashion? Finally, why would the engineer do that, especially with folk music?



Via SOS forum



Is a file really clipping? The only way to tell is by using a true–peak meter, such as this one in Steinberg’s Cubase.Is a file really clipping? The only way to tell is by using a true–peak meter, such as this one in Steinberg’s Cubase.SOS Technical Editor Hugh Robjohns replies: First, let me highlight a small but critically important difference: hitting 0dBFS is not synonymous with ‘clipping’ — it’s a perfectly legitimate situation to have a sample reaching 0dBFS, and a signal is only ‘clipped’ if a sample should have been allocated a higher quantisation value than was available. The only way to check the real situation is to use a ‘True Peak’ meter, as specified in the BS.1770 loudness metering recommendations — there are plenty of those from various plug–in developers. Other forms of ‘clip meter’ in your DAW may illuminate before the onset of clipping, or when there’s one or more samples at 0dBFS, and some may be user–calibrated — you need to be sure what your meter is actually telling you before you pay too much attention to the flashing red lights!



The fact that the material doesn’t sound clipped or distorted would suggest that it is, in fact, reaching 0dBFS by design (through a normalisation process or very precise limiter), but not actually being clipped. However, it’s also worth noting that while it’s possible to hear just a single sample clipping, with some material you won’t hear it even if it lasts more than 16 samples. It’s partly frequency dependent, but also dependent on the reconstructed waveform, which ordinary peak sample meters make no attempt to analyse (hence the need for a true-peak meter).



Are ‘peak–level’ mixes useful as references? Yes, of course they are: it’s the sound character of the mix that you’re referencing, so if the mix sounds good to you that’s all that matters. Note, though, that playing things back at the same loudness is essential if you’re to make meaningful comparisons.



There are, unfortunately, plenty of mixes that are genuinely clipped, which therefore distort to some degree. “Why would ‘they’ do that?” is a question I’ve been asking for decades! It’s technically unnecessary and ultimately destructive, but ‘art’ is often illogical, and pressure from the misguided ‘loudness wars’ lobby has encouraged or forced people to do daft things for decades, sadly.  

Published in SOS February 2015