How did they do it ? Sound spacialisation at mixing - Daft Punk & Agnes Obel

Question

Without being a sound engineer, I've always been impress by how real engineers manage to give a 3D feeling to a song. I've been wondering on how it is physically possible. Especially when I listen to the two following songs :
- Agnes Obel - Riverside
- Daft Punk - Veridis Quo (from 0:40 to 1:00).

Here are some hypothesis that I formulate on my own and I would love to have yours comments :
- I do not believe that a simple reverb could do such a good job.
- Could they simply duplicate the exact same voice track and shift it by a few milisec ?
- Does Agnes Obel managed to record so exactly the same voice twice and the mixing is done that one is set to the left ear the other to the right ear ?
- Is daft punk changing slowly the instrument from 0:40 1:00 to create this feeling ?
- Is daft punk playing with some kind of spectrum enhancement / enlargement to reach the 3D effect ?

Is there any other technique that is used and I've not been thinking of ?

Thank you for helping me to satisfy my curiosity ;)

Regards,
Vincent

Answer 1

Tracing back exactly the steps an engineer took in order to create that „magic“ mix will be quite difficult and it has a lot to do with accurate monitoring and especially with a ton of experience. I will try to give you some tips anyway:

The „3D Effekt“ of a stereo mix could be separated into the stereo panorama, produced by the L and R channel. Here, mid and Side processing comes to mind, which is the principle of mixing the side signal (S = 0.5*(L-R)) and the mid signal (M = 0.5*(L+R)) separate from each other. This enables you to exercise more control over the stereo panorama of your mix, or you might say the horizontal propagation of your mix. A lot of plugins offer this kind of processing, izotope and brainworks being the most active from the top of my mind. With MS processing, keep in mind that the side channel is only audible in stereo playback!

The second part of „3D mixing“ could be the „depth“ of the mix. The position of audio material on an axis moving away from you, orthogonal to the L and R axis. As you already mentioned, reverb does comes into play here. Many plugins offer the option to process early reflections and late reflections separately. Generally, a rule of thumb I use for reverb and distance would say: - the more early reflections and the less late reflections (or diffuse field) gain, the shorter the distance - the shorter the pre delay the shorter the distance - the more dry opposed to wet signal, the shorter the distance There are some other aspects of distance perception. The air absorption in real life causes small wavelengths to be absorbed while longer wavelengths persist while propagating through space. Therefore, a damping of high frequencies can amount to longer perceived distance. There is a great freeware plugin called proximity, which models some of these psychoacoustics related to distance perception. It’s very subtle and I use it for every mix to add depth.

You can never know what additional trickery they used in the daft punk production, but maybe these tips help a little bit anyway ;)

All the best, Simon

Answer 2

Space in sound has a lot to do with stereo separation - how distinct the left and right channel are, or how much infortmation the shared "middle channel" is carrying compared to the unique left and right channels are.

The haas effect (delaying one stereo channel by a few ms) is quite good at stereoization, but it audibly offsets the sound either to the left or the right, which is a pain to fix. Programmed drums are given room by running them through a subtle reverb that is either really quiet or has a very short decay time.

When algorighmic reverb isn't enough to make something sound distant, low pass/high shelf filters are used to reduce the high frequencies.

A good tool is convolution reverb, which is realistic by design, as it figures out the environment of a recorded impulse by processing the recording. You may have come across impulse response files (IRs, the aforementioned recordings) being used for guitar cabinet emulation, but their original purpose was to emulate reverberation in real environments.

There are a lot of freeware vsts (ozone imager, space knob, panagement) made for the purposes mentioned above. Play around!