With the coming of the new R128 standard of measuring audio a new method for measuring peaks and limiting also came along. This new method gives us the ability to look for peaks that are actually crossing that 0 dBFS line.
Sample peak meters
In the ‘old days’ we regarded 0 dBFS as the ‘end’ of our meter. Nothing could get any louder while the perceived audio clearly sounded louder to our ears as long as we keep pushing the limit.
The reason for this audible effect is that the analog signal actually is cut off early while your DA converter is still trying to finish the soundwave. Your meters won’t show the overload because you properly limited the audio at 0 dBFS.
But as the meters in your DAW are sample correct, they only look at the actual sampled values and see that the maximum value is followed by a second maximum value and so it regards that as being 0 dBFS. And so there’s no distortion going on as long as we keep everything under 0 dBFS.
Technically speaking your DAW is right in the sense that there are no ACTUAL sampled values over 0 dBFS, but in practice there are.
True peak meters
So with the new R128 standard they have also tried to tackle this problem. They invented intersampling meters (IS meters).
In short: At the moment of measuring we not only measure the actual sampled value and the next, but also IN BETWEEN the two sample moments. How many, depends on which meter you’ll be using, but it mostly is four times.
With this method we measure what each sound waves is actualy trying to do. So if a sine is passing 0 dBFS in a very steep manner, it probably will pass the ‘red line’ by a great deal before coming back to the next sampled value.
This way of measuring gives us a more accurate reading of the level of the maximum peaks. And that way it can happen that your accurately limited audiofile at 0dBFS still creates level OVER maximum.
In other words; R128 still regards that as creating distiortion, which it is (in fact anything hitting 0dBFS is distorted).
Even if you’d set your maximum at -0.3 dBFS it would cross that line. So the EBU standard prescribes that no signal must cross -1 dBFS, so the sine has a chance of not crossing 0 dBFS. Also it is advised to have an integrated level of -23 for broadcast purposes. Now I’ve written in another article that you could use a different mark for your integrated level(-16 dBFS would be nice). But the mark of -1 dBFS for your peaks seems like a good mark.
Meters in your DAW
Most of you will be using a native DAW. These DAWS do not have a IS metering. You can test for yourselves how that affects your system by playing a 12.000 Hz sine in a 48 kHz samplerate project.
The sampled values of this sine are completely in sync with the sample clock. So there is optimal quantization of the samples. You should see a steady value at your meter. Now take a proper R128 loudness meter and insert it at your master. This meter should show you a steady level also.
Now play a 12.000,1 Hz tone in the same project and have a look at your meter. It should show a value that varies. It varies because the values generated by your tonegen are not completely in sync with the samplerate. Although it still is as loud and about the same frequency. Now take a look again at your R128 meter, it shows a steady tone: this meter does IS measurements of the incoming audio and computes the actual signal level.
R128 is not only trying to help you get back your dynamics but is also helping you to keep your audio files distortion free.
Most of the DAWs these days come with their own R128 meters as per standard. Try to use them and make yourself comfortable with these new helpful tools.