stereo widening on individual tracks
when to widen and when to leave things narrow. a source-by-source guide to stereo widening vocals, guitars, synths, and drums with mono-compatible techniques.
four tools, one decision
you have four tools now. panning and level differences. M/S processing. Haas delay and allpass decorrelation. you understand how each creates stereo width and what it costs in mono. the question is what you actually do with them on a vocal, on a pad, on a kick drum.
this is where most guides stop. they teach the techniques and leave the decisions to you. the result is producers reaching for the same widener on everything. an orchestra does not put every instrument in the centre. the spatial arrangement IS the mix.
width is not a texture you add uniformly. it is a spatial decision you make per element, based on the role that element plays in your mix. a pad can be wide because it lives at the edges of the arrangement. a lead vocal should not be wide because it needs to anchor the center. a kick drum needs to be mono or the low end collapses.
key takeaway
the goal is not maximum width on every track. it is the right width on each track, so the elements create a coherent spatial image when they play together. narrow elements make wide elements sound wider by contrast.
the three questions
before you widen anything, ask three things.
does this source need to anchor the center?
lead vocals, bass, kick, snare: these elements define the middle of your stereo image. widening them weakens the center. a wide lead vocal sounds unfocused. a wide kick sounds like the low end is drifting. keep anchor elements narrow or mono.
does it have a mono compatibility constraint?
anything that will play on a phone speaker, a bluetooth device, or a club PA needs to survive mono summing. Haas delay creates comb filtering in mono. allpass decorrelation is dramatically safer. if mono compatibility matters (and in 2026, it almost always does), default to allpass.
does it already have recorded stereo content?
a synth pad with a built-in chorus, a pair of drum overheads, a room microphone recording: these sources arrive with stereo information already present. M/S processing lets you boost the existing Side content. allpass decorrelation or Haas delay works on mono sources where no stereo content exists. choosing the wrong tool for the wrong input wastes processing power and adds artifacts.
correlation checkpoint
check the correlation meter before and after widening any source. a reading of +1.0 means the signal is perfectly mono. zero means the left and right channels are completely unrelated. below zero, you have phase cancellation. for individual tracks, stay above +0.3. for the full mix bus, stay above +0.5.
vocals
your lead vocal anchors the center of your mix. widening it weakens that anchor. in almost every genre, the lead vocal should be mono or very close to it.
the perception of vocal width comes from what surrounds the vocal, not from the vocal itself. reverb returns panned left and right, a stereo delay with different timing on each side, background harmonies spread across the stereo field: these supporting elements create a sense of space around a centered lead. the lead stays focused. the arrangement breathes.
background vocals are different. harmonies, ad-libs, doubles: these can be widened aggressively. they serve the arrangement, not the spotlight. hard-pan genuine doubles left and right. use allpass decorrelation on a single background vocal to create synthetic width. stack and spread them to fill the sides.
a common mistake is widening the lead vocal to make it “bigger.” it does sound bigger on headphones. then you check on a phone speaker and the vocal has lost body, definition, and presence. the mono fold-down collapsed the very width you added.
tip
the widest-sounding vocal mix uses a narrow lead surrounded by wide supporting elements. reverb, delay, and harmonies create the spatial impression. the lead itself stays perfectly centered.
pads and synths
synthesizers are the most rewarding sources for stereo widening because they respond dramatically. this is not a coincidence. research on stereo perception shows that your ears are most sensitive to inter-channel phase differences in the mid-frequency range, roughly 400 to 2000 Hz.[^1] this is exactly where pads and synths carry their most energy.
already-stereo synths
many synth patches ship with built-in stereo: unison detune, chorus, stereo spread modes. before reaching for a widener, check the correlation meter. if the synth’s internal stereo already pushes correlation below +0.5, adding more width risks mono problems.
for already-stereo content, M/S processing is the natural choice. it lets you boost the existing Side content without creating new phase relationships. the stereo image expands within the boundaries the synth already established.
mono synths
a mono synth pad is the ideal candidate for allpass decorrelation. the signal has no existing stereo content, so decorrelation creates width from scratch. the mid-frequency peak of allpass decorrelation (around 600 to 800 Hz) aligns perfectly with the harmonic range of most pads.
bass synths
keep them mono. always. your ears cannot localize frequencies below about 80 Hz, so stereo bass provides no perceptual width. what it does provide is phase differences between left and right that cancel on mono playback and waste headroom. if a bass synth patch has built-in stereo content, sum it to mono before processing.
a bass crossover (mono below 100 to 200 Hz) before any stereo widening protects the low end while letting the midrange and highs expand. KERN WIDE’s FOCUS parameter does exactly this: it collapses existing stereo below the crossover frequency and only applies widening above it.
guitars
guitars present a simple decision: did you record two takes or one?
double-tracked guitars
if you recorded two genuinely separate performances of the same part, pan them hard left and right. this is the gold standard of guitar width. the natural performance differences between takes (timing, pitch, dynamics, tone) create rich, natural stereo content that no processor can replicate. no widener needed. no artifacts. perfect mono compatibility.
single takes
a single guitar recording is a mono source. you can widen it, but the results are fundamentally limited. allpass decorrelation duplicates the signal with frequency-dependent phase shifts. the problem is that both channels still contain the same spectral content. at high width settings, the artificial origin becomes audible: a phasey, processed quality that sounds very different from the natural richness of a double-tracked pair.
keep synthetic width below 50% on a single guitar take. subtle decorrelation adds enough spatial interest to lift the guitar out of dead-center without announcing itself as an effect.
key takeaway
double tracking beats any widener. if you have the option to record two takes, do that instead of processing a single take. the natural differences between performances create stereo content that sounds richer, more natural, and more mono-compatible than any synthetic widening.
drums
drums require the most careful stereo thinking because they span the entire frequency range: sub-frequency kick, midrange snare, broadband cymbals. widening everything uniformly creates problems. widening selectively creates a mix that sounds both punchy and spacious.
kick and sub
always mono. zero exceptions. a stereo kick wastes headroom and creates phase cancellation on every mono playback device. if your kick sample has stereo content baked in, sum it to mono. if your bass drum microphone pair is in stereo, collapse it.
overheads and rooms
these are where the legitimate stereo image of your drum kit lives. the overheads capture the natural spatial relationship between hi-hat (left or right), ride (opposite side), and the kit’s overall width. this is real recorded stereo content.
M/S processing is the natural tool here. you can boost the Side channel to widen the overhead image, or attenuate it to narrow the kit’s footprint. the stereo content is already there. you are just adjusting its level.
one critical detail: overhead microphones capture bass content from the kick drum and toms. that low-frequency stereo content does not help you (your ears cannot localize it), and it hurts you in mono. a bass crossover that collapses stereo below 150 to 250 Hz protects the low end of your overhead signal. this is exactly the problem that M/S processing and bass-crossover tools are designed to solve.
individual hits
snare, toms, and other close-miked sources sit in the center or at moderate pan positions. widening them rarely helps and often creates an unfocused, smeared image. keep close-miked drum elements centered or panned, not widened.
heads up
widening a drum bus without a bass crossover sends the low-frequency stereo content from overheads to both sides. on a phone speaker or bluetooth device, that bass image collapses and cancels. always use a mono-below crossover on widened drum content.
the mixing context
every track you widen reduces the correlation of your mix bus. a wide pad, a wide guitar, and a wide drum overhead might each sound great in solo. together, they can push the bus correlation below +0.3 and create a mix that sounds hollow or thin on mono devices.
track-level widening compounds
if you widen five elements at 60% width each, the combined bus is wider than any individual element. the width accumulates. before adding bus-level widening, check the correlation of the bus without it. if the bus is already below +0.5 from individual track widening, the tracks are doing the job.
leave room for contrast
not everything needs to be wide. narrow elements make wide elements sound wider by contrast. a mono vocal in the center makes the stereo reverb around it feel more expansive. a centered kick makes the wide overheads sound bigger. if everything is wide, nothing is wide.
the mono check
before you finish, fold the mix to mono. the mix should lose depth and spatial impression, but no instrument should disappear. if a source drops dramatically in level or changes tonally, its widening is creating too much phase cancellation. back off the width on that element or switch to a more mono-safe technique.[^2]
tip
A/B your widening with two checks: stereo bypass (does it sound wider?) and mono fold-down (does the mono version lose instruments?). if you pass both tests, you are in safe territory.
frequently asked questions
frequently asked questions
should lead vocals be widened or kept mono?
your lead vocal almost always works better mono or near-mono. a wide lead pushes the center of your mix outward and loses focus. use supporting elements to create width around it: background vocals panned wide, reverb returns in stereo, subtle room on the sides. the lead itself stays anchored. widen harmonies and backgrounds, not the main performance.
how do you widen a synth pad without phase issues?
use allpass decorrelation rather than Haas delay. allpass shifts phase without changing the magnitude spectrum, so the mono fold-down stays tonally intact. if your pad is already stereo from the synth itself, check the correlation before adding more width. many synth stereo modes are already near the safe limit. widening them further pushes correlation below +0.3 and causes audible cancellation on mono systems.
should kick and bass always be mono?
yes, for almost all music. your ears cannot localize frequencies below about 80 Hz, so stereo bass gives you no perceptual width. what it does give you is phase differences that cause cancellation on mono playback and headroom loss at the low end. keep kick, bass, and sub elements mono. if your kick sample has stereo width baked in, sum it to mono before processing.
how wide should a stereo guitar sound?
it depends on how you recorded it. a genuinely double-tracked guitar with two separate performances panned hard left and right sounds wide and natural. a single mono guitar through a widener sounds artificial above 50% width. if you only have one guitar take, use subtle allpass decorrelation and keep the width conservative. double tracking beats any widener.
can you use stereo widening on individual tracks and the mix bus at the same time?
you can, but track-level widening and bus-level widening compound each other. if you have heavily widened pads, guitars, and synths, the mix bus correlation is already well below +0.7 before any bus widening. adding more at the bus pushes into dangerous territory. the better approach: widen individual elements intentionally, then check the bus correlation before touching the master widener.
references
a note from the developer
when i built KERN WIDE, the hardest part was restraint. the algorithm can widen anything. 40 cascaded allpass filters with ERB-spaced center frequencies and a perceptual width curve tuned to how your ears actually respond to decorrelation. it sounds impressive when you crank it. the problem is that “cranked” is almost never the right answer.
the mixing decisions in this guide are the ones i use myself. lead vocal stays centered. pads and harmonies go wide. bass stays mono. drums get a crossover. the FOCUS parameter in WIDE exists specifically because i kept making the same mistake in my own mixes: widening a drum bus and forgetting that the overheads carried bass content into the stereo field.
the perceptual width curve peaks around 600 Hz because that is where listening tests showed the strongest decorrelation sensitivity. not because i decided 600 Hz was important, but because ears decided it. the algorithm follows the research. the mixing decisions follow the ears.
if your widening approach looks different from mine, or if a genre i did not mention has different rules, jonas@kernaudio.io. these guides get better with more perspectives.
try it yourself
KERN WIDE: psychoacoustic stereo expansion that keeps mono compatibility. $29, no iLok, no subscription.
built on this research
WIDE applies this science in real time. five knobs. $29. no iLok.