Do I need to change gain if merging from stereo to mono?

dhjdhjdhj's icon
Max Patch
Copy patch and select New From Clipboard in Max.

I have a patcher that takes two separate inputs, by default goes through 1.0 gain (allowing me to adjust gain if I want) and then goes to two outputs of a dac. Here's the section of the patcher in question

Now suppose I need to merge those two inputs (left and right) so as to go out through a single channel rather than two channels. Do I have to reduce the initial gain from each input before I connect them to a single input port on a dac?

bkshepard's icon

If both signals are scaled to 1.0, then there is a strong likelihood that the combined signals will produce an output above 1.0 when their peaks line up. Most DAWs that allow you to combine a stereo signal into mono give you the option of reducing the signal by 3dB in the process to prevent clipping. I've always made it a practice to do the same thing in Max by using [*~ 0.5] on either the two channels before summing, or on the summed signal before it gets to the [dac~].

Peter Castine's icon

Just a note: [*~ 0.5] would be -6 dB. For -3 dB (which is the typical adjustment) you need [*~ 0.71]. Ask a pocket calculator for the exact value of sqrt(2) if you want more precision (don't forget, about 6 digits is as good as it gets for Max).

The -3 dB guideline may depend a little bit on your material, but it's a standard rule-of-thumb.

bkshepard's icon

Oh, I always thought the DAW was doing a -3dBm conversion, which would be [*~ 0.5]. Is that not the case? Is it dB SPL?

dhjdhjdhj's icon

Thanks for the responses --- the goal is that I am sending a stereo pair to channels 1/2 at FoH such that the two faders are at 0db (say) and then I temporarily sum my stereo pair to a single channel that I now send out on channel 1, the FoH engineer should not have to adjust the channel 1 fader at all.

José Andrés Medina Ochoa's icon

what about if I want to join 3 signals?

Source Audio's icon

-3db per channel

Roman Thilenius's icon

might sound strange... but what he said.

the more channels, the higher the chance that peak values appear at the same time, but it also gets more likely with many channels that they do not contain full scale values, so you always go by -3db.

it is not mathematically reasoned, it is just a working convention.