Your browser is out-of-date!

Update your browser to view this website correctly. Update my browser now


With Regards to Nigel Tufnel

Foti says improvements in RMS detection offer advances in creating perceived loudness

The author is president of Omnia Audio.

Nigel Tufnel might’ve been thinking about broadcast audio processing when he eloquently described his theory how “this one goes to 11!”

Yes, it’s that hilarious scene from the movie “Spinal Tap.” Nigel’s comments about going to 11 may appear tongue-in-cheek, but we now have the means upon which to make his prognostication true.

Standing the test of time, radio stations constantly seek a magical method to empower their signal in order to capture and hold listeners. In a word, they try to be louder than everyone else. Except now, the industry has reached a point of diminishing returns, as radio stations have pushed their existing equipment to the limits.

Theories abound about how to create perceived loudness. The trick is how to achieve this, yet minimize the annoyance and distortion commonly associated with competitive processing. This article offers insight to a new method, available now, and it is on the air in most major markets. Without a need for rocket science, the basic concept is not new, but the means by which it is accomplished are.

A pleasing secret

In general, audio processors funnel wide dynamic range audio down to the precision peak-controlled signal needed for transmission.

This is accomplished through a series of processing functions such as compression, limiting and distortion-managed clipping. Each provides its own function in order to obtain maximum sonic benefit. Simply, the compressor (least aggressive sounding) feeds a peak limiter (more aggressive sounding), and then the distortion-managed clipper, which is the most aggressive element.

Common practice has been to utilize the limiter and clipper for creating loudness, and the AGC to balance the spectrum. While this does work, it brings along associated intermodulation (IMD) distortion, which results in sonic annoyance to the listener.

Recent efforts to reduce and in some cases eliminate IMD have improved processing systems, but that’s not the ultimate answer. Even though lowering IMD helps, limiters and clippers are still aggressive-sounding, no matter what.

The secret to generating consistent and pleasing competitive loudness is to be found in the AGC or compressor section. This is not novel or new. What is novel is how this section generates its internal control signal that enables consistent sounding and non-aggressive audio.

Research shows the human ear is an RMS (root means square) detector. This is how we judge the loudness of one signal to another. The ear basically ignores a change in short-term peak level, but it locks onto a change in the “averaged” level over a period of time. Limiters employed in broadcast processors are peak-responding, and react over a much shorter duration. In order to increase the RMS level through limiting means, we need to ingest more limiting to increase the perceived RMS level. This is counter-productive, as considerable amounts of limiting are required. While the RMS level eventually will increase, so will the perceived level of IMD, and this is where processing artifacts are generated.


Effecting a change in RMS level is done in the AGC/compressor section. RMS-based compressors are not new. They have been deployed before; and based upon how the detector operates, they had a few sonic challenges to overcome, until now.

In prior implementations, difficulties were based upon the single time constant used to calculate the RMS level. This works great when measuring steady-state signals. Audio signals, on the other hand, cover a wide range of spectra; and the averaging function of the RMS detector is not consistent in level detection over this wider range.

The result is inconsistent level control, as well as occurrences where the audio signal will sound abnormally weak, and usually non-musical.

This problem can be minimized through the use of multiband AGC/compression, but it still occurs. Actually, in the multiband case, a new and additional problem arises. When RMS detection errors occur, they now affect the sonic level, as well as the tonal balance of the spectrum. The result is a combination of incorrect levels and EQ, which is very non-musical and a double negative.

All is not lost, however. The magic lies in the RMS detector. As mentioned, a single averaging function is used. New development, in which the averaging function is performed through analysis of the signal’s incoming density and rate of level change, allows the RMS detection to be accomplished in a natural and musical manner. All levels are adjusted properly across the audio spectrum. This enables the AGC function to operate in either a wideband or multiband configuration. When used in the multiband mode, the resultant spectrum is consistent in level and frequency balance.

Too many bands?

Why is this so important? Now, the mechanism is in place to set RMS (loudness sensitivity) levels to where they need to be, and keep them there. Also, spectral balance is achieved consistently over a broad range of content … music, live speech, commercials, station sweepers, etc.

Now, here’s the added benefit. Since the less aggressive-sounding sections, the WB-AGC and multiband compressors, have set the RMS (loudness) levels and EQ to the desired target range correctly, the following peak limiters and distortion-managed clippers are then set for moderate operation.

This yields less activity in the multiband limiter (reduced IMD), and the desired amount of distortion-managed clipping is consistent at all times!

This sheds some light on another misnomer about processing for level control, EQ balance and loudness. It has little to do with the number of processing bands. It has more to do with the control algorithms employed in the gain management sections. Using six or more bands in a limiter does not offer improved quality, EQ, detail and loudness. Too many limiter bands driven deep into processing generate dense, smashed and annoying sound.

With this new RMS implementation, the result is a sonic signature that is extremely clean, level and EQ consistent, as well as competitively loud.

This innovative method has been tested over recent years, as we brought our latest product to life. It was compared to our prior offerings, and the improvement in performance is easily apparent. I trust Nigel would be proud.

Comment on this or any article. Write to [email protected].