The ultimate goal of any public address system is to be heard. However, because the amount of underlying background noise in a retail store, restaurant, hotel or transportation space varies as people come and go, audio engineers find it difficult to set an appropriate volume level suitable for both peak and quiet times. While the engineer can certainly add manual control to adjust the volume, it’s also possible to employ an effect known as Ambient Noise Compensation (ANC) to automate volume adjustment.
I’ve been looking closer at ANC, addressing how the algorithm works and how engineers can use it to control sound levels in public spaces. In my first post, I looked at how to select ambient microphones, the mics that listen to the noise level of the space. In this second post, I’ll address how to use the signal from those microphones to manage volume levels, using ANC.
Adjusting the ANC Settings
The ANC algorithm works by using microphones installed throughout a space to capture ambient noise. The signals from these microphones are then input into a digital signal processor (DSP), like the BSS Soundweb London Series or dbx ZonePRO Series. The DSP then adjusts the volume of the program audio (the background music, PA announcements, etc. that are going over the sound system) based on how loud the ambient noise is in the space. When the noise level in the space increases, the ANC algorithm increases the program audio as well.
You may recognize this as a dynamics effect—something that I’ve written about previously—and that is essentially how ANC works. Instead of monitoring the program audio, the algorithm monitors the audio from the ambient mic; but otherwise, ANC works similarly to other dynamics effects and has threshold, ratio, attack and release controls. However, as a specialized dynamics effect, the ratio and threshold settings work slightly different than in other dynamics effects. If you aren’t familiar with how dynamics effects work, I suggest reading this article before continuing here.
Like with other dynamics effects, the threshold is the point at which the algorithm engages. Once the volume in the space (as measured by the ambient mic) goes above the threshold, the algorithm turns the program audio up, so people in the space can hear it. Volume increases according to the ratio (in other words, the higher the ratio, the more the algorithm turns the volume up once it goes above the threshold).
Attack and release work similarly to the way they do in other dynamics effects. Attack determines how quickly the system turns the volume up when the noise level goes above the threshold; the release determines how quickly it turns the volume back down once the sound level in the space goes back down below the threshold. However, unlike a compressor, where the attack is measured in milliseconds (so it can be extremely responsive), the ANC attack and release are measured in full seconds, with options of up to 50 seconds before adjusting sound. This allows the audio engineer to ensure the volume does not increase in response to a short burst of sound (like someone yelling).
ANC offers two additional settings that other dynamics effects typically don’t: minimum and maximum settings. This essentially allows the audio designer to establish a “box” for how loud and how quiet the sound can be. The algorithm won’t drop the volume below the set minimum, nor will it push the volume above the set maximum. The maximum level, in particular, is important to ensuring the system doesn’t get so loud that the speakers get damaged. It also helps prevent “runaway audio” (something we’ll address later in this article).
Gap vs. Non-Gap ANC
There are two types of ANC: gap and non-gap. So far, I’ve been discussing non-gap ANC. The distinction refers to when the algorithm is engaged. Non-gap ANC is always measuring and adjusting volume, while gap ANC only measures during “gaps” between the program audio.
Non-gap ANC is available in BSS Soundweb London as well as dbx ZonePRO. Since this version of the effect is always engaged, it is good for situations with background music or other sound that is always (or usually) playing. An example of a good application for non-gap ANC is a restaurant or bar. In these situations, there may be constant background music playing, but there are also “table-ready” announcements going to specific zones as well. Since there is no gap in program audio (the music is constantly playing), the effect must sample the noise while program audio is present. For more information about how non-gap ANC works in the dbx ZonePRO, check out this video.
The BSS Soundweb London Series also offers a gap ANC algorithm, which is only engaged during the “gaps” between program audio. Gap ANC is designed for situations where there are bursts of audio from the PA system (announcements, etc.) with gaps of quiet time. For example, in a museum, there are gaps between the paging announcements, where there is no sound from the speakers. Since the amount of people in a museum can vary greatly depending on the time of day, the ANC effect can measure ambient noise during the quiet periods to ensure the level is correct when there is a need to make an announcement. By only engaging when the program audio is quiet, the system is able to more accurately measure the sound level in the room, because it isn’t picking up the program audio coming from the speakers.
Adjusting Gap ANC Settings
Since gap ANC only measures ambient noise during gaps, there are some important differences to the settings. To adjust how the algorithm identifies a gap, the effect adds two more settings: gap threshold and gap time control.
The gap threshold identifies the point at which the program volume is low enough to be considered a gap. In other words, the gap threshold helps the system identify when the program audio is “quiet.” When the program audio falls below the gap threshold, the algorithm begins monitoring the ambient noise level.
Gap time control, on the other hand, adjusts how much time the program audio must be quiet before it starts monitoring the ambient noise. This is particularly useful in highly reverberant environments. In these situations, a long gap time can keep the system from measuring the room echo of the announcement rather than the actual ambient noise level.
To accommodate for these new settings, there is no ratio setting on gap ANC (the algorithm works with a preset 1:1 ratio). It also has a single “speed” setting for the ANC algorithm instead of a separate attack and release. Like non-gap ANC, it does have minimum and maximum settings.
It’s important to note that the gap settings monitor the program audio, while the rest of the gap ANC algorithm monitors the ambient mic. The gap settings monitor where the gaps are in the program audio, and then tell the algorithm when to start adjusting volume based on the ambient mic.
Handling Runaway Audio in Non-Gap ANC
Gap ANC is useful for situations without constant background audio. If the system is playing background music, the gaps are not frequent enough to properly measure and compensate for the noise level in the room. However, because non-gap ANC is always monitoring room noise, it also has the potential to pick up the program audio coming from the speakers. This can create the problem I mentioned earlier that is sometimes referred to as “runaway audio.”
If the ANC algorithm picks up the program audio coming from the speakers, it considers it “room noise” and turns up the volume. It then registers that as increased room noise and turns the volume up more. This cycle causes the volume to increase exponentially, “running away,” as it were. This is one of the reasons why it is important to have a good maximum volume setting, to prevent runaway audio from causing speaker damage. A limiter effect placed after the ANC effect in the audio signal chain is also a good suggestion as a secondary backup.
Using ANC with AEC
If you have limited options for ambient mic placement, but you need to use non-gap ANC for the application, the system may be at risk of runaway audio. An alternate suggestion in this case is to add acoustic echo cancelation (AEC). For an overview on how AEC works, check out this article.
While AEC is designed to prevent echo on audio calls, the technology can also be used to prevent runaway audio. AEC works by monitoring program audio and then filtering that audio out of another signal. For this application, this means that you can use AEC to filter the program audio out of the ambient mic signal, providing a suitable reference of the noise level for everything other than the program audio. Only the ambient noise is passed to the algorithm, so it can provide a more accurate response.
An example of when this would be useful is for train platform announcements. In this application, the program is not continuous, but gap ANC would be ineffective. Trains may arrive while announcements are being made, increasing the ambient noise and drowning out the announcements while the effect is “off” (remember, gap ANC only works when there are no announcements taking place).
To ensure announcements are heard over any sudden volume changes, non-gap would probably be used. However, since the volume from the speakers would potentially be turned up very loud, very fast to compensate for the train noise, the system would run a very real risk of runaway audio. Unless the ambient mics are placed in a location that won’t pick up the sound from the train noise, using ANC with AEC allows the system to compensate for train noise while preventing runaway audio.
This is just the tip of the iceberg when it comes to what’s possible with ANC and other processing capabilities in HARMAN Professional Solutions audio systems. To learn more, you can sign up for training from HARMAN Professional University, where there are online and classroom courses available.
Do you have tips on how to use ANC successfully? Share your insights in the comments.