Mix on Nay
Target: unstable_break · bratty placement to stabilize M1.5 across the passaggio
A Suede vocal companion
Suede Sing
Your voice has a signature. Suede Sing reads it — range, register transitions, breath support, resonance, vibrato — every session, on-device. The result is your TonePrint, and a practice plan built from it.
Built on real research
What used to require a sound booth, a Praat script, and a graduate degree now runs in the background of every session you record — voice-lab-grade analysis, on your device, ready when you sing.
What gets measured
Inside your TonePrint
Every sustained note becomes a row of objective measurements. Glance for the headline number, scroll for the spectral detail.
Mixed voice (M1.5)
M1↔M2 transition zones detected via H1−H2 differential inversion across the F0 sweep. Reports passaggio entry frequency, exit frequency, and the spectral signature at each.
H1−H2 differential
dB(H1) − dB(H2) per sustained note. Negative or near-zero = M1 (chest); ≥ +6 dB = M2 (falsetto); intermediate = mixed.
Spectral tilt
Linear-regression slope of dB amplitudes across H1–H4 in dB/octave. Steeper negative slope ↔ more falsetto-like glottal source.
Cepstral peak prominence (CPP)
Prominence of the cepstral peak at the F0 quefrency, dB above the linear baseline of the surrounding cepstral window. High CPP ↔ clean, periodic phonation.
Harmonics-to-noise ratio (HNR)
10·log₁₀(harmonic-energy / non-harmonic-energy) in dB. Quantifies what your teacher means by “support.”
Singer’s formant cluster
Energy ratio in the 2.0–3.0 kHz band, normalized against total spectral energy. The acoustic signature of professionally projected sound.
Laryngeal height (inferred)
Approximated via spectral centroid drift and F1/F0 ratio shift across phonation. Surfaces the compensation patterns ear-training cannot quantify.
F0 stability
Standard deviation of fundamental frequency across the stable window, expressed in cents. Distinguishes sustained tone from wobble, drift, and tremor.
Pitch accuracy
Mean cents deviation from equal temperament (A4 = 440 Hz) across the sustained window. Resolution: ~5 cents (parabolic-interpolated YIN).
Vibrato characterization
F0 modulation rate (Hz), depth (cents peak-to-peak), regularity (CV %), onset latency (ms). Distinguishes singer’s vibrato from tremor and wobble.
Range mapping
Chest ceiling, mix span, head onset, head ceiling, absolute span. MIDI semitones with per-register confidence bands, accumulated across every session.
Practice that fits
Drills selected by your data
The right exercises for the gap you actually have. The recommendation engine reads your diagnostic tags, intersects them with your current working range, and rotates so you don’t see the same drill twice. Most singers compensate without knowing it — pulling chest above where the laryngeal mechanism wants to flip. Suede Sing surfaces the compensation in dB and prescribes the retraining.
Octave Jump — F4 to F5
Target: low_passaggio · smooths H1−H2 inversion at register boundary
Chest Pull, Step 2
Target: weak_chest_top · vowel modification for safe chest extension
Watch your voice grow
Every metric, charted across every session
The progress you’ve felt becomes the progress you can prove. Every session adds a data point; the trajectory tells the story. Persisted on-device, exportable as a JSON + WAV bundle when you’re ready to share with a teacher.
Bring it to your coach
Or meet a new one, fully briefed.
Export your TonePrint, your recordings, and your tracked metrics to the vocal coach you already work with — or book a Zoom session with a Suede-vetted teacher who’s already read your data. A coach with your acoustic history skips the diagnostic phase entirely. Three sessions of “let me hear you” collapse to zero.
Current build
Where Suede Sing is today
Foundation complete
Audio capture, spectral analysis, register classifiers, range engine, recommendation engine, subscription, and notifications — all built and tested.
Interface in build
Onboarding, calibration, capture, review, history, settings, paywall — in active development.
App Store launch
Submitted to Apple — waiting on review. Drop your email and we’ll write you the moment it’s live.