Split a track into mid, side, left, and right

Mid/side in practice: hear the width, then change it

Every stereo file secretly carries two other channels. The mid — left plus right — is everything mixed to the center: lead vocal, kick, snare, bass. The side — left minus right — is everything that makes the mix wide: reverbs, doubles, stereo synths, the ghosts.

This page splits any file into those parts, shows where the width lives by frequency, and then lets you put the file back together with the balance changed.

§How to split and rebalance a track with mid/side

1. Drop a stereo fileYou get four playable channels — mid, side, left, right — plus the full track on the transport.
2. Listen to the side firstThat is the width in isolation: reverb tails, doubles, wide pads. The classic sample-digging channel.
3. Read the width spectrumOpen Width spectrum to see which frequency bands carry the stereo image.
4. Rebalance or mono the lowsSet mid/side gains in Rebalance, or pick a crossover in Bass mono to fold the low end to center.
5. Render the stereo WAVThe recombined file downloads as uncompressed 16-bit stereo.

01Two lines of algebra, zero loss

Mid is (L + R) ÷ 2 and side is (L − R) ÷ 2; going back, left is mid + side and right is mid − side. Encode and decode are exact inverses, so splitting a file this way loses nothing — it is a rotation of the same two channels, not a processing step. That reversibility is what makes M/S safe to experiment with: unity gains give you back the original, bit for bit.

02Where the width lives

Width is not spread evenly across the spectrum. Most produced music keeps the lows nearly mono — energy and playback systems both demand it — and pushes stereo information into the mids and highs where reverbs, doublings, and cymbals live. The Width spectrum mode measures side-versus-mid energy per frequency band and draws your file's actual profile.

Read it like a map before surgery: a file whose width peaks at 2–8 kHz will respond to a side boost with air and sparkle; a file carrying side energy below 150 Hz has a mono-compatibility problem you can now see, and the Bass mono mode is its answer.

03Rebalancing: the honest width knob

The Rebalance mode applies plain gain to each channel before recombining: mid gain moves the center, side gain moves the width. Worked example: +3 dB of side multiplies the side amplitude by 1.41 — which doubles its energy — so a file whose side held 20 percent of total energy comes back with roughly a third. Audibly wider, with the vocal and bass untouched because they never lived in the side.

This is what most 'stereo widener' plugins are underneath, honestly labeled. If the recombined peaks would clip, the tool trims the output and tells you by how much, so the render never distorts silently.

04Mono below the crossover

Low frequencies punish width hardest: club systems sum the subs to mono, vinyl cutting physically cannot hold wide bass in the groove, and anti-phase low end cancels into mush on any single-speaker playback. The fix is old and standard — make the mix mono below a crossover. The Bass mono mode high-passes the side channel at your chosen frequency (60–300 Hz), so below it left equals right exactly.

The proof is printed next to the render: correlation measured on the lows before and after. After the fold it reads +1.00 by construction — that is the algebra doing the work, not a promise.

05Geometry, not separation

M/S manipulates positions, and positions are shared. Boost the side and you raise the reverb, the tape hiss, and the cymbal spill together — the algebra cannot tell a lush double from noise, because both simply live off-center. Cut the mid hoping to duck a vocal and the kick, snare, and bass duck with it.

And width bought with side gain is invisible to mono listeners by definition: the side channel is exactly what a mono sum discards. Physics grants no exceptions.

06M/S shows where things sit. Stems hand them to you.

The side channel is a beloved sample-digging trick precisely because it almost isolates things — almost. It is geometry: whatever shares a position comes along. SongStrip in the studio does the version that actually isolates: GPU stem separation trained to know what a voice, a kit, and a bassline are, returning clean stems from any stereo file. Dig the side here for ghosts; split the song there when you need the instrument itself.