Checklist Items Under Test: 6.33: For conferencing mode, at the 65dB SPL listening level, be able to demonstrate full duplex operation, with no reports of echo or “speech trails” (as detected from the far end).
Reasoning: There is a growing trend in the AV industry to combine elements of conference room AV, broadcast and live event engineering in today’s multipurpose rooms. One of these combinations includes integrating audio mixing consoles with DSP-based matrix mixers. By doing so, the designer hopes to give the operator the flexibility and manual control of a console, with the ease of use of a DSP-based system. However, a number of criteria must be met to do so successfully, both on the operation side of things, as well as the performance side.
The Story: AVR was hired to commission a multipurpose room that had two modes of operation: Automatic and Operator. In Automatic mode, all the system microphones were controlled by a large DSP mixer. We tested that first, and it went really well. In Operator mode, a mixing console was deployed in the room so an operator could run the board in the space. For presentations, this mode worked really well. The operator had access to an incredible amount of audio controls.
However, it did not perform well when a conference was added to the presentation. The microphones were all sent to one input in the DSP, and the echo canceller could not adequately cancel the reference from 16 microphones on one channel. The entire audio system had to be reworked, and fast, because the VIPs preferred having an operator at the board for their high-end meetings.
In other words, the Automatic mode, even though it worked well, was not acceptable. It cost the client tens of thousands of dollars because the design was not reviewed properly.
It sounds great on paper: “We’ll give the operators a mixing console so they can have control over the event, but we’ll also integrate the board to a [insert favorite DSP-based matrix mixer here] as a security blanket.” It sounds easy enough to implement, as well. The drawing is simple: Just connect the two mixers with a few cables (or one network audio cable) and, hopefully, someone will sort things out in the magical, black box DSP, right?
However, once it gets into the field, that’s where things start to get dicey. Which device is responsible for microphone levels to the echo reference? How many microphone signals can have echoes canceled on a channel without affecting performance (one or five or 16)? How much flexibility does the operator actually need at the mixing console? What path does a microphone signal actually take in the system? This solution gets very hairy very quickly.
There are two main methods of integrating the two mixers as I see it, and the difference is which device is directly connected to the microphones. The first method has all the microphone signals go to the DSP first and then pass through to the mixing console. Although this requires a significant amount of inputs and outputs on the DSP (an input for each microphone, an output for each microphone, plus all the other system requirements), it has the best conferencing performance because the DSP can cancel audio from each microphone individually. However, it runs the risk of taking a lot of audio controls away from the operator (possibly leading to incredible frustration). The DSP programmer is tempted to mix the system mostly in the DSP realm and only leave level controls and EQ for the operator.
The second method has the microphones going to the console first, and then to the DSP in various groups. The cost of the DSP required for this scenario is reduced, but so is system performance possibly. However, this option typically gives the operators more flexibility at the console, with less functionality happening behind the scenes in the DSP engine. If the operators are used to running events completely from the board, having system routing happening automatically in the DSP may be confusing and limiting.
Both methods can work, but they require a lot of design work and forethought. A drawing is not adequate to describe the successful implementation of this scenario. Each audio path between the mixers must be thought out and described carefully. The solution must include information about how the operators want to run the events. Do they want to simply control levels and EQ? Do they want control over different audio destinations (Speech In Room, Various Room Zones, Audio Call Tx, Video Call Tx, Recording Channels, etc.)? Given the audio path solution, can the DSP effectively handle the echo cancellation? If the interface between mixers is analog, is each path properly labeled in the drawing? If the interface between mixers is digital, does the design include a table detailing all the required audio paths? Further, do all the operators possess the skills to proficiently run an event with this solution? As I said, showing the two mixers connected with a Cat5 cable ain’t cuttin’ the mustard.
As a commissioning agent, I am more concerned with the “what” than the “how.” If the system performs according to the specification (the “what”), I am not too concerned with what’s going on underneath the hood (the “how”*). However, as a design reviewer, in order to confirm that the system will perform as designed, I am very interested in the “what” and the “how.” During the design review, I need to make sure all the necessary questions are asked, answered and included in the design package to pass on to the next member of the project team.
(*Actually, if I’m being honest, I’m always interested in the “how.” My favorite part of my job is observing the genius solutions and new ways of doing things created by those in our industry. I should say I’m not “contractually concerned” with the “how” during commissioning…but I digress.) n