Acoustic Echo Cancellation (AEC)
AEC requires an AHM-64 or an AHM-32 running V1.10 (AHM-64)/V1.20 (AHM-32) or higher to be outfitted with the M-AHM-64/M-AHM-32 Processing Expansion Module and provides 12 channels of echo cancellation.
ⓘ AEC is not currently available for the AHM-16
Concept
Acoustic Echo Cancellation (AEC) is commonly used to mitigate the sound of a far-end caller on a conference or video call hearing themselves come back through the near-end. The near-end mics (local microphones in a meeting room, for instance) are being constantly adjusted and filtered to remove what the AEC sees as far-end (a Teams or Zoom call, as example).
Any Input in AHM (sourced from anywhere) can be sent into the AEC.
Insert A on an Input is designated for near-end mics (local microphones).
ⓘ There must not be non-linear processing (Gate) before the Near End reaches the AEC
Insert B on an Input is designated for the far-end mics (remote caller).
ⓘ There must not be non-linear processing (Compressor) between the Far End source and the speakers in the room
A Sound Reinforcement (SR) Output is provided for local sound reinforcement (e.g., in larger rooms or theatres where attendees need voice lift) post-AEC but prior to any talk-state gating or comfort noise. This allows for appropriate local reinforcement of the same Inputs that are going into the AEC. Labelled AEC SR Returns, the SR Outputs of each AEC instance are available for routing under Assign > XPoints > AEC SR Returns/Zone.
Assignment
Click on Insert B on the Zoom or Teams computer and click Assign AEC (Far End). Select an available AEC Channel. Any AEC Channel already in use will be greyed out.
The Insert block will populate with the controls for the Far and Near Ends.
Click the Near End Assign field to choose the Input that will act as one of the local in-room mics. This can also be done by going directly to Insert A on the Input and clicking Assign AEC (Near End).
Controls
Far End
| Assign | Change the Input that is acting as the Far End (remote caller) |
| Trim | Manual or Automatic Gain Control. For best performance, aim for 0dB (meter on the right side of the Insert B block) |
| Gain | Available when Manual Trim is set. +/-18dB available |
Near End
| Adaptive Filter | Switch + Scale 1-10. Enables and controls the speed of the adaptive filter. Set to lowest value that gives acceptable performance for application |
| Echo Reduction | Switch + Scale 1-10. Enables and controls non-linear processing to further reduce echo. Set to lowest value that gives acceptable performance. A higher value can cause far-end caller to hear distortion |
| Noise Reduction | Switch + Scale 1-10. Enables and controls non-linear processing to further reduce background noise from the local room. Set to the lowest value that gives acceptable performance. A higher value can cause far-end caller to hear distortion |
| Far-Only Ducker | Toggles the attenuation of the Near End signal when only the Far End is talking |
| Comfort Noise | Controls the level of unobtrusive noise sent to Far End to ensure the caller never hears complete silence (which can give the impression the connection has dropped) |
Multi-Mic Setup
In many situations, multiple microphones are used to cover a large seating area in the Near End, or there are multiple speakers each with their own microphone. In these cases, we can set up the AEC to be Mult-Mic.
Simply click Add AEC (Multi-Mic) to add more AEC Channels to this AEC instance. Each new local microphone will require another AEC Channel for proper filtering and processing. Toggle the Near End section of the AEC block(s) to view the assignment and controls per microphone.
ⓘ There is not currently a way to have multiple Far Ends associated with a single Near End
Feedback Suppression (FBS)
AHM’s Feedback Suppression (FBS) system offers 8 instances of 25 Bands to eliminate frequencies that get “caught” in a sound reinforcement scenario where a microphone is hearing itself come back through the loudspeakers, typically causing a “squealing” or otherwise unpleasant, loud sound. FBS detects the prominent and persistent frequencies and precisely applies narrow filters to reduce their level while changing the overall sound as little as possible.
Currently, FBS is only available for AHM-64 and AHM-32 that are outfitted with the Processing Expansion module. More on this Slot and how to configure it can be found in the Processing Expansion section of this guide.
FBS can be inserted onto any Zone through the Insert block found in Channels > Zones. Clicking Assign Feedback Suppressor will prompt a selection of any available FBS instance. Once chosen, a grid is displayed with a dB (y-axis, -30dB to 0dB) over frequency (x-axis, 20Hz-20kHz). The overall response of the EQ is shown in yellow when it is active, and grey when FBS is bypassed.
Click Start Detecting to allow FBS to begin listening for problematic frequencies. As any feedback is detected, a notch filter (Band) will be implemented.
When Bands are implemented, their Band Number will turn red (live) or blue (fixed). Any unused bands will be heavily darkened.
Selecting a Band under the Band Select section of the block will colour the filter red in the grid, and relevant information to that Band is shown in the Band Info area. In this area, it will indicate whether the band is Fixed or Live, what its Frequency is set to (ranging from 20Hz to 20kHz), as well as its adjustable Q (ranging from 8.00 to 640), and the gain being removed (ranging from 0dB to -18dB). Reset a particular band with the Reset Band X button in this area.
Assign how many fixed Bands to implement using Number of Fixed Bands – click and drag or double-click to type in a value between 0 and 25. A value of 0 means the FBS will always be adjusting and listening to make adjustments. If feedback is not detected after a certain length of time, the filters are lessened and eventually removed. A value of 25 means all the Bands will be fixed once they have been assigned to a specific frequency. Feel free to use the amount of each that is appropriate for the project. All 8 instances of FBS will continue to have 25 Bands available regardless of what the other FBS instances are doing.
Response is how quickly FBS responds to perceived feedback. Ranges from 30ms to 1.2 seconds.
Recovery (in seconds) is how long after the offensive frequency drops below the FBS’s threshold that the attenuation is removed. Ranges from 1 second to 1 minute.
Bias Q is the amount of Q applied to offset or compensate any Band. Ranges from 0.2 to 5.5.
Bias Gain is the amount (in dB) of gain applied to offset or compensate any Band. Ranges from -6dB to +15dB.
Reset All Bands will remove any set filters.
Unassign FBS X will unassign the insert and free up the FBS instance for use elsewhere.
ⓘ A Reset FBS function is found on Allen & Heath’s IP Controllers and AHM’s local controllers (Front Panel, GPIO, etc.) to give end users the ability to refresh the FBS and have it begin its process again
ⓘ When inside of a Project, many tabs include Unit-specific options. Change the Focus Unit by selecting the dropdown menu of the same name in the lower left of System Manager’s main window