close X
EN / RU
close X

Thanks!

Your application has been accepted. We will call you back soon

registration

What Changes for the Sound Engineer with Source Intelligence?

all news
1
Feedback is familiar to every sound engineer. A microphone picks up the PA, the amplified signal returns to the microphone, the loop closes, and at some point the unmistakable howl appears.

The conventional ways of controlling it are well established: careful loudspeaker and microphone placement, directivity management, EQ, level control, and, where appropriate, gates and dynamics processing. All of this depends on experience, critical listening, and a solid understanding of the acoustics involved.
L-Acoustics Source Intelligence introduces a different approach by addressing the problem closer to its source.

Running in real time on L-ISA Processor II, Source Intelligence separates the human voice from the other sound captured by a vocal microphone: spill from the PA, drums and instruments on stage, audience noise, and room reflections. It uses machine-learning-based source separation, with unwanted signal attenuation of up to 40 dB.
It is important to understand what the technology does not do. It does not correct pitch, reshape the singer’s performance, or generate a new voice. If the artist sings a note differently, Source Intelligence leaves that performance intact. Its role is to isolate the voice from its acoustic surroundings before the engineer begins shaping the channel at the console.

That leads to an interesting professional question:
If an algorithm can remove up to 40 dB of unwanted sound and significantly increase gain before feedback, does system tuning become too easy?
Probably not. It changes where the engineer’s attention goes.

Source Intelligence does not override the laws of acoustics. A poorly designed reinforcement system will still behave like a poorly designed system. Incorrect directivity, excessive stage energy, unsuitable microphone placement, or errors in array design and tuning will continue to affect the result.
What changes is the quality of the starting point.

Traditionally, part of every soundcheck is spent managing compromise. The engineer wants to preserve the natural character of the voice while reducing drum spill, stage noise, and leakage from the main system. The more aggressively EQ, gating, and dynamics are used, the greater the risk of affecting the openness and natural behaviour of the vocal.

Source Intelligence moves part of that corrective work earlier in the signal path. Instead of treating the consequences, the engineer receives a signal in which the voice is already much more clearly separated from its surroundings.

That means less time spent repairing the channel and more time available for the artistic decisions that define the mix.

This becomes particularly relevant on modern stages, where performers are no longer confined behind the main PA. Thrusts extend deep into the audience, stages can be circular, and vocalists move through areas that would traditionally have been considered acoustically challenging.

On the 5 Seconds of Summer tour, for example, four vocalists performed on a large ring-shaped stage alongside a loud acoustic drum kit. Source Intelligence was used across all four vocal channels to help control spill from the stage and the sound reinforcement system.

The technology is designed specifically for live applications. Latency is below 8.5 ms, four independent processing instances are available, and signal transport can be handled via MADI, Milan-AVB, or AES.

So perhaps the more useful question is not whether engineers will forget how to tune a system.
It is this:
What will distinguish a great engineer when part of the routine corrective work is handled by an algorithm?

System design, understanding the interaction between stage and audience, recognising balance, and making artistic decisions are likely to matter even more. Source Intelligence does not replace those skills. It gives engineers more room to use them.
On paper, the concept is easy to understand. In practice, this is one of those technologies that is much more revealing when heard than when described.
Take a microphone. Move closer to the PA. Add stage spill. Switch the processing in and out several times.

That is why, after L-Acoustics Rental Day, we left this conversation open. Source Intelligence can be heard and tested in our Immersive Lab in Tashkent.
The best way to understand where the algorithm ends and the sound engineer begins is to try it for yourself.