What is a true peak meter?
A true peak meter displays the absolute peak of an audio’s waveform as it will be heard in the analogue realm through speakers.
How Do True Peaks Affect Music?
Most DAWs show levels in the digital mixer using a Sample Peak Programme Meter (SPPM). This type of meter only shows the peak sample value and doesn't represent the ‘True Peak’ of the audio. The meters that we see in our DAWs make us assume that 0 dBFS (decibels full scale) on our output is our absolute limit. So, as long as we don’t exceed 0dBFS our audio will sound great…right?! We seem to have forgotten about the digital to analogue conversion that happens when we hear our audio through speakers.
All music created digitally in a DAW must be converted back to analogue before we can hear it. As part of this conversion a reconstruction filter is applied to round off the stepped digital audio signal. This gives us a smooth listening experience. These filters can cause slight changes in the levels of the audio. This can be a problem for the signals that are close to 0dBFS and can cause clipping. A high-end digital to analogue converter has headroom to compensate for this issue, but cheap speakers won’t be able to compensate for these inter-sample peaks. This means your mix won’t sound distorted in the studio, but your peaks might become clipped when played through a cheap Hi-Fi.
The music industry doesn't have a widely accepted practice to standardise the peak level of audio. Conversely, the TV and film industry have very strict guidelines regarding the loudness and levels of audio. Initiatives like ‘Mastered for iTunes’ have introduced a process whereby the mastering engineer delivers audio in a format that avoids inter-sample peak issues by increasing the headroom (space between the loudest peak of the music and 0dBFS). iTunes converts the final master to a lossy ‘AAC’ format. During this conversion process, filters are applied to reduce the information in the file to make downloads and streaming quicker for their users. Similarly to the digital to analogue conversion this process can cause fluctuations in the levels of the signal, which can result in the final lossy codec being distorted.
So inter-sample peaks mean that the audio with a peak of 0dBFS might distort in the analogue realm… So how can I avoid inter-sample peak distortion?
An immediate solution to the issue is to use a quality True Peak meter to make sure your audio won’t distort when converted from digital to analogue. This will give you the information you need to leave the correct amount of headroom between the peaks of your master and 0dBFS. It’s worth noting that not all true peak meters are created equal. This is because detecting true peaks requires the audio to be converted to a higher sample rate and then low pass filtered to get back an approximation of the original audio. According to the maths (as is ever the case in ‘Digital Signal Processing’) better approximations require more CPU cycles and so different true peak meters will have different accuracy/CPU usage trade-offs.
Mastering Audio Using True Peak
When mastering audio, it’s common practice to set the limiter at 0dBFS and push the audio hard into it. The goal is usually to get the track sounding as LOUD as possible to compete with all the other very loud tracks in the charts. This approach will inevitably cause the audio to clip on many playback systems due to the digital to analogue conversion. It’s known that the majority of chart hits during the ‘Loudness Wars’ years have a true peak around +1dBTP (decibels true peak) and up to +3dBTP (that’s quite a lot of distortion!). Some people argue that these inter-sample peaks are so unnoticeable that they don’t deserve our attention. However, inter-sample peaks are distorting the audio we hear. Would it not be better to master audio with a slightly lower peak and simply remove this distortion? A good true peak meter will give you an accurate reading of your audios peak level. By mastering to 0dBTP you’ll give your audience the best possible listening experience.
