All signal processing systems function within natural boundaries. Analogue gear performs within limits set by the power supply voltage and the circuit components. Any signal passing through an analogue mixing desk may have an amplitude value ranging between the low level noise floor and the breaking point of saturation. Digital systems function within a set of boundaries as well, imposed by quantizing specifications. In the case of a digital system, a signal’s minimum and maximum value will be constrained by its bit-depth (i.e. word length)
In order to investigate the effect of any audio processor on the dynamics of an audio signal, it’s important to explain some fundamental concepts
- Dynamic Range: This is the ratio between the maximum and minimum possible value of a signal, depending on the qualities of the system. It is a common term in audio and sound engineering and one that all professionals are familiar with. The concept of dynamic range is closely related to that of SNR.
- Signal-to-noise ratio (SNR): This is a measure that compares the level of the desired signal, against the level of noise, interference or any unwanted artifacts. The theoretical dynamic range of a system is the maximum theoretical SNR of a signal. A 24-bit digital audio signal can take values between -144 dBFS (minimum) and 0 dBFS (maximum). The maximum achievable SNR is 144 dB.
- Headroom: Strictly speaking, ‘headroom’ is defined as the amount by which the maximum possible level in a signal exceeds a given reference level. In other words, it is a safety zone, aiming to ensure that audio output is undistorted. For example, most mixing engineers will work in the box with levels ranging from -9 to -12 dBFS, effectively having 9 to 12 dB of headroom until the signal hits the ceiling (0 dBFS).
- Noise: The term ‘noise’ does not only describe a system’s natural shortcomings, but is a general term applied to any unwanted interference present in the signal path. Going back to our SNR definition, the implications are obvious: having a signal well above the quantization noise floor, does not necessarily mean an acceptable SNR. Anybody who grew up listening to metal or punk and had his mother tell him to “turn down that awful noise” is already well-versed in the subjective nature of the concept of noise.
In light of the above, let’s examine the case of drum recordings, in particular the recording captured by a tom microphone. The desired signal would be the sound of the tom, with the noise floor being the amount of leakage.
In the above example, the track is peaking at around -6 dBFS. The leakage has an average RMS of -33 dBFS. The tom hit would be at about -21 dBFS, giving us an SNR of 12 dB. Using equalization or compression on the captured signal, will more than likely have a further detrimental effect on the SNR. Within a mixing context, a low SNR would be translated as an absence of clarity and definition. Furthermore, a high noise floor means that the sound engineer is unable to take advantage of the system’s dynamic range, since raising the signal volume would also apply to the inherent noise. Drumatom aims to offer a solution to this problem: the noise floor is reduced without an impact to the energy of the desired signal. Let’s consider the previous example. The drum recordings are loaded into Drumatom and the group is analyzed. By selecting the tom track and applying processing, the user can reduce leakage level by several dBs. In fact, even a touch of the ‘Focus’ control has immediate effect on the noise floor (indicated on the suppression meter). This the same tom track, processed with Drumatom:
Here’s a time domain comparison between the two tracks:
And a frequency domain comparison:
The processed track has a noise floor at about -45 dBFS. That means that leakage has been reduced - and SNR has been improved - by 12 dB. This degree of improvement can be a big deal in a realistic mixing situation, especially considering the additive effect of mixing multiple processed tracks.
To sum up, as the control is turned clockwise:
- The noise floor is reduced.
- The signal’s SNR is improved.
- The desired portion of the signal gains an increase in dynamics.
Applying an equivalent amount of processing in two or three key channels of a group can have a beneficial effect on the group’s dynamics, resulting in a subjective sense of ‘openness’ or ‘space’ in the sound. Here is a similar comparison for a complete group:
As before, the red waveform represents the unprocessed group.
In realistic everyday scenarios, dealing with these concepts boils down to making small improvements that add up. Balanced decisions will yield a few dBs of additional dynamic range in some key tracks, further ‘polishing’ the sound and making a big difference in the final result. It should be noted that overprocessing a track is never a good idea: precision and moderation have been the cornerstones of successful mixing throughout the years.
For more information on drumatom, you can also watch some drumatom related videos!