Burnish Vocal Chain

Eight stages run in one window, in signal order. IN is trim, high-pass, gate, breath control and a noise floor. TUNE has a Classic path, the hard-tune sound at 0.1 ms, and a Natural path that corrects without moving your vowels, with key, scale, speed, humanize, throat and a note keyboard that sets which notes are legal targets. SHAPE is five bands plus a per-band adaptive de-harsh that cuts resonant excess rather than brightness, and a program-adaptive split-band de-esser. COMP is FET into Opto. COLOR is tape, tube and present character with an air shelf. DOUBLE is a detune-decorrelated doubler that folds down to mono under 3 dB. SPACE is delay and reverb, both ducked from the vocal. LEVEL is a feedforward BS.1770 short-term loudness rider, converging within ±0.5 LU, then a safety limiter. Drop a reference record on the REFERENCE well: an acapella, or a full mix, which is isolated first and labeled reduced confidence. The analysis extracts a 40-band K-weighted tonal curve, dynamics density, a sibilance profile, a space estimate and a tune style, fits interpretable rules, and shows each stage what it did and why. Match Amount scales the whole fit from 0 to 150 percent. It is offline, background and fully local; nothing leaves the machine. The screen is the take. The bright line is your voice after correction, the faint steel line under it is the pitch you sang, and the lanes are the legal notes, named down the right edge, with gain reduction and short-term loudness above the trace. Mint sparks mark de-essed esses, the line splits magenta while the doubler runs, and the halo dims as the compressors work. Each station tile carries a small meter of that stage's work; select one and the deck's six knobs relight as that stage's main controls, with its own scope on the glass. Every deck value and every continuous value on the Design screen can be typed in its own unit: Hz, dB, milliseconds, seconds, cents, ratios or percent. TRACK runs zero-lookahead dynamics on the Classic path at 0.1 ms total, four samples at 48 kHz, so a singer hears the finished chain in the headphones. MIX runs full quality and reports its latency exactly; the header names the mode and the figure at all times, 12.5 ms with the Natural path. Sixty-six factory chains each carry a measured target profile, so loading one arms Adapt to my voice, and twelve Learn chains isolate a single stage each. 131 parameters, add-only. BYPASS is latency-aligned true bypass and MATCH A/B is loudness-matched, and both are click-free, as is every stage toggle and the Track to Mix change. Burnish Vocal Chain is a lead-vocal chain: mono or stereo in, stereo out, with polyphonic material out of scope by design. The reference proposes settings so your voice sits the way the record sits; nothing can make one voice sound identical to another, and Burnish will not pretend otherwise. Built and used at a working recording studio.