Stereo width explained: beyond the Haas delay

Stereo width is the sum of three independent things: how correlated the left and right channels are, how much energy sits in the Side channel relative to the Mid, and whether one side arrives a few milliseconds before the other. A 20 ms Haas delay only touches the third. That is why dropping it on every wide element produces the same flavour of width every time, and why the kick disappears the moment a phone speaker sums the mix to mono.

What "width" is in signal terms

A perfectly mono signal has identical content in both channels. Correlation is +1, the Side channel (L minus R) is silent, and there is no arrival-time difference. Move any one of those three variables and the brain registers spatial spread.

• Correlation. Two channels carrying different content (not just different levels) read as wide. Different content can come from a real second performance, from modulation that constantly reshuffles the relationship, or from reverb tails that diverge L/R. A correlation meter shows the running average: +1 is fully mono, 0 is fully decorrelated, negative is out of phase. Hugh Robjohns, writing for Sound on Sound, points out that "it was an essential tool for cutting vinyl records, to ensure the stylus wouldn't be thrown out of the groove by any strong out-of-phase components." Same meter, same job today: it tells you what your mix will do when a streaming codec or a Bluetooth speaker sums it.
• Mid/Side balance. The Side channel is what cancels in mono. Boosting it makes the mix wider; cutting it makes it more mono-safe. M/S does not invent new stereo content, it only scales what is there.
• Arrival-time difference. Identical content arriving at one ear a few milliseconds before the other is interpreted as spatial position. This is the Haas zone, roughly 2 to 35 ms. Outside it, you get either phase colouration (below 2 ms) or audible echo (above 35 ms).

Width comes from one or more of those three variables changing. Reach for the technique that moves the variable you need to move.

When each tool is the right one

Panning moves a source to a position. A hard-panned mono synth sits at one point on the soundstage with perfect mono compatibility. Do this first, before any width processing, because anything you layer on top of a well-panned arrangement works less hard.

Double-tracking is the only technique that produces genuinely decorrelated, tonally independent width. Two real performances panned hard left and right do not cancel in mono because they were never the same signal. Ken Lewis, writing for the Universal Audio blog, puts the principle bluntly: "The more variation you can put into similar sounds, the wider and bigger the stereo impact will be." Matt Wallace takes it further, telling Tape Op that he leaves the duller of two doubled guitars EQ-uncorrected, noting that "The dull guitar part is reinforcing the other side, so that it gives it a bigger stereo feel." Tonal contrast between sides widens the image without needing the pans to go any further out.

This is also why quadruple-tracking four identical takes does not sound bigger than two well-differentiated ones. Correlated copies add level. Decorrelated performances add width.

Haas / short-delay widening is the right tool for sustained, harmonically simple sources where you want the sensation of spread without re-recording anything: pads, room mics, an organ, sometimes a strummed acoustic. It is the wrong tool for anything transient-heavy (snares, hats, percussive guitars), anything below 200 Hz (kick, bass), and lead vocals in dense mixes, because the comb filter that appears on mono summing eats the body of the sound.

Stereo chorus does what a static Haas delay does, except the delay time is being modulated by an LFO, so the comb filter sweeps rather than sits. That swept character sounds less artificial on close-listened sources like electric pianos and lead synths, and it collapses to a gentle tremolo in mono instead of cancelling.

Reverb on a stereo return widens a centre-placed dry source by adding decorrelated tail energy around it. The dry signal stays anchored. This is the right move when a lead vocal or a snare needs more space but has to hold the centre position.

M/S width control is corrective, not generative. Use it on a finished mix or on the master bus to scale stereo content that is already there. A 2 to 4 dB Side boost above 6 kHz opens the top of a mix without bloating the low mids. Pushing the Side knob on a mostly-mono mix will not rescue it; there is nothing in the Side channel to scale up.

How to dial in a Haas delay without wrecking mono

If a Haas delay is genuinely the right call for the source, the dial-in is short:

1. Duplicate the track, pan original and copy hard L/R.
2. Set the delay on one side between 10 and 25 ms. 15 ms is a reasonable starting point. Below 5 ms the result skews directional and phasey; above 35 ms it stops fusing and becomes an audible slap.
3. High-pass the delayed copy at 250 to 400 Hz. Low-frequency phase cancellation in mono is the single biggest mono-compatibility problem with this technique, and you almost never need the low end of the delayed side to feel wide.
4. Sum to mono and listen. If the source thins, hollows, or partially disappears, the delay is too long, the high-pass is too low, or the source is too transient-rich for this technique. Pick a different tool rather than fighting it.
5. Check the correlation meter on the channel. A running average that sits around 0 is fine. Consistently negative means something will cancel on a phone speaker.

A working order of operations

Pan first. Double-track the elements that carry the song's width: rhythm guitars, key backing vocals, the hook synth. Use Haas only on sustained, harmonically simple sources that survive a mono check with a 300 Hz high-pass on the delayed side. Use stereo chorus where you want the spread to breathe rather than sit. Use a stereo reverb return when the dry source has to stay centre. Save M/S for the mix bus and master, where you are scaling stereo content rather than inventing it, and watch the correlation meter the whole way through.

If every wide element in a mix is wide for the same reason, the mix will sound like one stereo trick repeated. Different sources want different widening mechanisms, and the choice of mechanism is where stereo image stops being a knob and starts being an arrangement decision.