NoiseAssist Simplified

Ever wonder what’s going on behind the scenes when you use your NoiseAssist plugin for 8-Series or MixPre II Series? Let’s take a brief dive into the process of noise suppression. In this series of diagrams, the black sine wave represents the wanted dialog signal, and the orange lines represent the unwanted background noise.¹

Figure 1: This is an example of how an incoming signal from a microphone may look if it were viewed on an oscilloscope or in a DAW. The dialog and the noise are intermixed and the noise rides on top of the dialog waveform. This is how the unprocessed signal appears at the input stage of the NoiseAssist algorithm.

Figure 2: NoiseAssist separates the incoming signal into two different signals: the wanted signal (dialog) in black, and the unwanted signal (background noise) in orange. This real-time separating of the signal is at the heart of NoiseAssist. NoiseAssist uses multi-band frequency, level, historical and statistical info to do hundreds of millions of calculations per second to accurately separate the foreground signal from the background signal. These calculations are made possible by the powerful FPGAs within the 8-Series and MixPre II Series recorders.

Figure 3: The background noise signal is intelligently suppressed independently from the foreground signal. This suppression is controlled by the user’s NA setting in the menu, given in dBs. This is the average amount of suppression across the entire frequency spectrum.

Figure 4: The background signal and the foreground signals are recombined to form a similar signal to the original input signal. The final signal is very similar to the original incoming signal in figure 1 — the foreground dialog signal is unaffected — but the background signal is now at a lower level. This entire process from start to finish takes place in real-time (1 ms).

¹ These diagrams have been simplified for illustrative purposes. In this scenario, the dialog is a relatively low-frequency signal, and the orange noise is a far higher frequency. While this is sometimes the case, it is by no means the case most of the time.