Zero-latency monitoring. What is it? Why should you want it?
Computers are very convenient for recording, and comparatively inexpensive. The only problem is that they are not designed for audio. So although they are great for word processing and surfing the Internet, they have a little problem when it comes to sound.
And that problem is 'latency'. When you put a signal into a computer, it takes a while to process before it comes out again. I'd call it a delay, but latency is the word that has been accepted for this meaning.
The reason for the delay is that computers are not optimized for signals. They like data they can pick up, process and put down at any time that suits their convenience. Quite quickly you would hope, but a computer has no sense of urgency.
But signals are totally dependent on time. They need to be processed millisecond by millisecond. The trouble is that computers just don't do that.
So the answer is to use a 'buffer'. A buffer is a temporary storage area for audio data. So when the computer feels like it, it can process some audio faster than real time. And then while it is attending to some other task it has in hand, the buffer can fill up some more.
The buffer therefore averages out the rate of flow of data. The drawback is latency. And the more processing that is required, the bigger the buffer needs to be, and the longer the delay.
So to put this into a real-life context, you are recording yourself singing, with headphones on. You are listening to the output of the computer, via the audio interface of course.
But there is a noticeable delay on the output. Depending on your system it might vary from noticeable to completely unworkable.
There is such a thing as zero-latency monitoring, which is provided by many audio interfaces.
What happens is that signal is tapped off in the analog domain within the interface and routed directly to your headphones, mixed with the other signals coming from the computer - your previously recorded backing tracks.
So you can hear yourself exactly in time, and I do mean exactly. You will sing several milliseconds out of time with the backing tracks, but in this context latency is unnoticeable.
The only drawbacks to zero latency monitoring are firstly that you have to mute your vocal signal coming back from the computer. Some softwares make this easier than others.
The other is that there is no processing on the monitor signal. Many people prefer to sing with a bit of EQ, compression and reverb.
Even so, zero-latency monitoring is a great invention. One day computers will be so fast that latency isn't a problem. We won't miss zero-latency monitoring when it is no longer required. But for now, it's essential.