How Mid-Side Processing Works in Audio Engineering

Mid-Side (M/S) processing is one of the most powerful — and often misunderstood — techniques in audio engineering. It's also the core technology behind free browser-based vocal removal tools like 7By.in.

The Stereo Field Explained

A standard stereo audio file has two channels: Left (L) and Right (R). When you listen through headphones or speakers, the relative level of each channel creates the impression of sound coming from different positions — left, right, or centre.

In most commercial music:

• Lead vocals are panned to the centre (equal in L and R)
• Kick drum and bass are centred
• Guitars, keyboards, synths are panned left and right
• Reverb and spatial effects tend to be wide and stereo

The M/S Matrix

Mid-Side processing converts the traditional L/R representation into two new signals:

The Mid signal captures everything that is identical in both channels — i.e., everything centred in the mix. For most songs, this is primarily the lead vocal.

The Side signal captures everything that differs between channels — i.e., everything panned off-centre. This is primarily instruments, backing tracks, reverb tails, and stereo effects.

Converting Back to L/R

To return from M/S back to stereo:

This round-trip is perfectly lossless — no audio information is lost in the conversion.

How Vocal Removal Uses M/S

The trick is simple: to remove vocals, discard the Mid channel and reconstruct stereo using only the Side channel:

To isolate vocals, keep only the Mid channel:

Professional Uses of M/S Processing

Beyond vocal removal, M/S processing is used throughout professional audio:

• Mastering — compress or EQ the Mid and Side channels independently. Boost mid-range in the Mid for vocal clarity; cut low-end in the Side to tighten the low end.
• Width control — reduce the Side level to narrow a too-wide mix; increase it to widen a mono-ish recording
• M/S microphone recording — a cardioid mic (Mid) + figure-8 mic (Side) captures a stereo field that can be adjusted in post
• Noise reduction — unwanted noise often appears in both channels equally (Mid). Targeted reduction in the Mid can reduce hiss without affecting stereo content.

Limitations

M/S separation isn't magic. It fails when:

• Bass guitar and kick drum are also centred (they get removed too)
• Vocals are doubled and panned (common in modern pop — they appear in the Side)
• Heavy reverb on the vocals bleeds into the Side signal
• The mix is mono (L = R, so Side = 0 and the whole track is "Mid")

How 7By.in Uses M/S Processing

7By.in's Vocal Remover and Stem Splitter implement M/S matrix processing using the Web Audio API's OfflineAudioContext. This allows the computation to happen entirely in your browser — no server required, no uploads, no privacy compromise.