Reverse tutorial - advice sought (updated 150322 with example mp3)

I’ve been trying to get my head around the art of mixing, and have lately been trying to learn about placement in the stereo field. I have widening plugins which can work nicely on pads, but I have been exploring instruments such as a piano.

At any distance, this is a point source, so although the sound in toto might be placed at any point in the stereo field, it comes from one direction (other than reverb or sound bouncing off walls) - unless you are standing , say, immediately behind the pianist. Even if it’s widened, all notes are spread around equally.

So I tried splitting the input midi up by note value and assigning it to one of 5 channels spread across the field, going from low left to high right, with middle C, well, in the middle. It seems to produce an interesting effect with some pieces (e.g. a solo piano).

So a question to all the non-newbs / pros out there, is this done routinely, or is it considered unspeakably naff because the piano is a point source ?

Recording Piano is difficult, my favourite process was recording James Blunt and his keyboardist in London. We set up the usual stereo mic on top of the cabinet and experimented with two ribbon mics behind the piano at the edges. That gave us the bottom end warmth and width without the notes sounding detached. From there it was the usual imaging in which using a haas effect worked really well (duplicating channel and delaying the right hand side by 15-25ms).
It all depends on what style of piano you are using. If its hip hop / trap style hard hitting piano then its about a narrower field and hard FET compressions to give it snap and attack.
If you are discussing purely processing an exisiting library then less is more as they are all recorded very well (similar techniques to what I described above).
Hope this helps but happy to help out more if you want to be specific.


in case of vst virtual instruments, like some of mine, I see they have a Mixer option with “spacial” sliders
On the other hand, if you look at the video about the making-of those VSTs you see multiple mics & effects were used to record samples

so do I understand well that in this case any added effect like the one proposed by @yorkeman is redundant?

No, because it’s quite different, and hence my question. If a piano was miked with 5 mikes spaced across its width, you would get a width effect, but the sound would be recorded by all 5 mikes - just at different amplitudes.

In my potty scheme, a set of notes is assigned to a given position in the stereo field. I have 5 instances of a piano VST, each panned to a different position, and any note is only fed to one VST, and so those notes do not sound in any other position … zero spill between channels. [I.e. C1 would only be in the very left channel and C7 only in the extreme right]

So the sound it produces is very different to stereo miking]

If you listen in headphones, since you only have two ears spill occurs, but some notes only appear in one ear - quite unlike real life.

Hence, it produces an entirely artificial sound, but I found for some pieces IMHO interesting. This is why I posed my question, and @davide’s answer was instructive, as the sound my layout produces can’t be reproduced by miking, nor ‘width’ VST’s.

I’ll post a small example.

Wow !

1 Like

I think its important to differentiate ‘space’ and ‘depth’ when producing. Space generally refers to how close or far something is in the mix according to mic positions (more or less room mics for example) or moving one instrument away from the other (left /right balance) or the decay of a chosen reverb (room/hall). Depth is more about depth of field and moving up, down and around using imaging techniques. The last five years has become ‘how wide can you go’ as opposed to ‘how low can you go’. Mid / Side techniques has further facilitated that in being able to treat the sides (what appears only in the far left or far right) as opposed to what appears in the mids (equally in both speakers) separately via EQ, Compression or any other effect. I guess I am saying there is no right or wrong for any specific recording and production techniques but understanding the tools available to you and their intention. From there its about implementation and what works for you. Well, there is one rule laid out by Joe Meek - ‘If it sounds right, it is right!’


Now can you produce a little song with the 2 settings, so I can understand the difference?
if my ear are smart enough :crazy_face: :rofl:

Ok, I’ve just done the file with the effects rather than without, as it’s pretty evident how it differs from normal stereo width. You will hear the very low notes go far left and the high ones go far right, but they do move around from L to R. Chords can be split by panning as well - you will probably detect this.
The set up looks like this

Important ! you need to listen on headphones to get the full effect <<<<

As @davide noted, the judgement is by the ears, and I have managed to get a few interesting effects.
The file is at stereo piano experiments – Galaxies and beyond …


The page you are looking for is temporarily unavailable.
Please try again later.
2c922c67e97becf896db269ad4ed6cf5 2263010e57d3713edcb6f42a41e1dfd3 2b8515c1974eb2924c85c5f1f075ce0a

URGH, it worked one nanoseond after…
later indeed :crazy_face: :rofl:

Interesting, really interesting

A casual listener anyway could think there are just 5 pianists tickling different ivories

Now, I am asking to myself is some famous musician used this trick already…

I’m still fiddling with the process to get the right splits and pans (it will vary for each piece), but (to my ears) I’ve had some useful effects.

Maybe @davide might comment in the morning whether the idea is ever used (it maybe routine for all I know). I suspect that it’s possibly too ‘artificial’ (but so is vocoding …)


I’m sure it has been used. Whether any particular artist uses it regularly I can’t answer but I’m sure a few do. It sounds lovely by the way, real nice. If I was asked to produce it I would be adding some weight to the bottom notes as they anchor the piece. I would keep them more centred and allow the melodic lines to play off them, I’d also make sure there’s some nice bottom end to the lower notes. Try Hass over the lot, it would really have a great sense of depth. Duplicate the whole thing. One lot hard left and one lot hard right. On the right add any delay, makes sure it’s 100% wet and feedback is low (approx 15%) and delay by 15 ms.
Great work I think it sounds lovely.


‘Moving into the box’ in early 2021 I have come to appreciate that it’s one thing to get a sequence of notes, but quite another in making them sound good. It’s a massive learning curve, and of course I don’t even aspire to reach professional standards, but just to be a bit better each time.
Thank you for your incredibly helpful comments. I can now hear the mismatch in level balance and flaws in placement in the field, before even starting to add any fairly dust.
Scaler has been an amazing tool in progressing my (limited) knowledge of music theory; to me, the really great thing is the immediacy of relating sounds to changes in chords etc.
I shall try and implement the changes you suggested !


It seems to me that you always do an exhaustive analysis of things. Thanks for sharing your research. I personally loved your version of the reverbs and pans in 5 pianos

1 Like

Check out this free instrument and look into the binaural recording technique used. When played back with headphones this gives that realistic stereo behind the piano sound.

Yes it is an interesting question. Pianos don’t move, and their sound does not move from one ear to another in the physical world unless there are two (or more instruments) that are in conversation. But really that is just a constraint of physical instruments. Playing the low notes on the left and high notes on the right is almost like hearing the piece from the pianist’s position. If you like the sound, and I do like your sample, why not? It is always an artistic risk of course.

In Cubase Pro 12, and probably other DAWS, there are a panning functions which you can automate for a single track, which could also give you the panning effect using only one track.

I have been using iZotope Neutron advanced lately and it has a feature called Visual Mixer that allows the movement of instruments around in the soundscape, both laterally and to and from the audience, to create effects of width and relative volume. One can simply create several copies of a track and delete all but the parts you want to hear from individual copies of the instrument. Moving the copies of the instruments around in Visual Mixer is very easy and you can hear the results immediately. Visual Mixer is used with Izotope relay on each track. I think this is still a free download, and it is quite a lot of fun to use, especially with multiple (and different) instruments.

Thanks for the useful comments here.
Of course, the sound is entirely artificial, but I did produce some interesting examples.

I did try with several copies of the original, but I used a velocity overlay to define the spatial splits. I used the chain effect simply because it was quicker (an non-destructive) to adjust the boundaries and move things around (plus rather than have 5 distinct pans, there were overlaps) than edit the files.

However, I’ll look again at the things I could do with Neutron; thanks for the post !