RTM Audio THROAT GOAT [MAC]

• Product: THROAT GOAT
• Developer: RTM Audio
• Version: 1.0.2
• Format: VST3, AAX, AU
• Requirements: macOS 10.15 or later
• Source: rtmaudio.com/throat

THROAT GOAT is a formant and vocal-tract character plugin from RTM Audio that reshapes SHIFT, GENDER, TILT, and FOCUS using a fixed-hop STFT true-envelope warp, leaving pitch and timing untouched by construction rather than correction. PROTECT guards consonants and sibilants during large formant moves, MOTION animates formants or tilt over time, and GRAB, STEAL, and MORPH transfer a captured voice’s timbre onto another source. It runs as AU, VST3, or AAX on macOS and Windows. It answers searches for a formant shifter that doesn’t touch pitch at all.

Key Takeaway

STEAL activates once a target voice’s timbre has been captured into a GRAB slot, blending the current source toward that captured character without touching its pitch or timing. It displaces manual formant-matching by ear across separate plugins. Only two capture slots exist at once, so a third distinct identity means recapturing over one already stored. Producers wanting subtle character shifts get more predictable results from SHIFT and GENDER alone.

Bins Don’t Move, Formants Do

THROAT GOAT tracks the vocal tract’s spectral envelope independently of the singer’s pitch using a fixed-hop STFT with true-envelope estimation, then warps that envelope at the point SHIFT, GENDER, TILT, and FOCUS are set to rather than touching the underlying pitch bins. Because the hop size stays constant and no bin repositioning happens, the pitch and timing of the original performance carry through unchanged regardless of how far the formant controls are pushed.

SHIFT moves the formant frequencies up or down across a ±12-semitone range, independent of pitch — pushed down it reads as a bigger, darker throat, pushed up a smaller, brighter one, with the sung note staying exactly where it was performed. GENDER scales the formant positions across a ±100% male-to-female range in one control, moving toward a different vocal identity rather than adjusting brightness or size individually.

TILT tips the whole spectrum by up to ±6 dB per octave, shaping brightness as a spectral slope rather than a fixed-frequency EQ boost, so the brightness change follows wherever the voice’s own energy sits rather than a static band. FOCUS tightens or loosens the resonant peaks themselves, sharpening or softening how defined each formant reads without moving its frequency.

THROAT GOAT changes formant character only — it doesn’t shift pitch at all, so a part that needs both a different note and a different vocal-tract character needs a separate pitch-shifting tool run alongside it. Producers reshaping an existing take’s character without touching its melody get the direct use case; producers needing an actual pitch change alongside formant work add a second tool to the chain.

Consonants Sit Still While Vowels Move

PROTECT guards consonants and sibilants from the formant warp while vowels continue moving with SHIFT and GENDER, keeping diction intact at settings that would otherwise blur or lisp through hard consonant sounds. The control defaults to 75%, a starting point RTM Audio states most sources want left where it starts rather than adjusted down.

At low PROTECT settings, a large SHIFT or GENDER move processes consonants the same way it processes vowels, which is where formant-shifting tools typically start to lisp or smear sibilants at extreme settings; raising PROTECT keeps those transient, high-frequency consonant sounds closer to their original character while the vowel content underneath continues to carry the shifted formant character.

PROTECT applies as one global amount across the whole signal rather than as a per-consonant or frequency-targeted control, so there’s no way to protect sibilants specifically while leaving other consonant types fully exposed to the warp, or the reverse.

Large SHIFT or GENDER moves aimed at a full identity change get the most benefit from keeping PROTECT near its default; subtle character moves that barely touch the formants have less need for PROTECT doing much work in the first place.

Motion Picks One Lane

MOTION modulates either the formant position or the TILT setting on its own, running from 0.05 Hz up to 12 Hz with a depth control from 0 to 100%, moving the target parameter around its set position rather than requiring manual automation to create movement.

At the slow end of its range, MOTION produces a formant or brightness drift that reads as a voice breathing or shifting character gradually over a phrase; pushed toward the fast end near 12 Hz, the same modulation moves quickly enough to stop sounding like a natural vocal quality and start reading as an unnatural, non-human texture instead.

MOTION modulates one target at a time — either the formants or TILT — rather than running independent modulation on both simultaneously, so animating formant position and spectral tilt on two different rates in the same patch isn’t available from MOTION alone.

Slow MOTION settings suit a performance that needs to feel alive and breathing without manual automation; producers wanting simultaneous independent movement on both formants and tilt need to automate one of the two by hand instead of relying on MOTION for both at once.

Traveling Between Two Captured Voices

GRAB captures the vocal-tract envelope of whatever voice is playing into the plugin at the moment it’s triggered, storing that captured formant signature into one of two available slots rather than applying it immediately. With two slots available, a session can hold one captured timbre as a starting reference and a second as a target to move toward.

STEAL blends the current source’s formant envelope toward whichever captured slot is active, with the blend amount controlling how far the source moves — low STEAL amounts shift the voice toward a family resemblance with the captured timbre, while high STEAL amounts move it far enough to read as a different voice performing the same take.

MORPH travels continuously between the two captured slots rather than blending toward just one, so automating MORPH across a vocal line changes which captured identity the voice moves toward over the course of that line while pitch and timing stay exactly where the original performance put them.

Only two capture slots are available at a time, so building a morph across three or more distinct captured identities means recapturing over one of the existing two slots. Producers building a single, targeted timbre transfer get full use from one GRAB and a STEAL amount; producers automating an evolving identity change across a phrase get the most from MORPH riding between both captured slots.

A Null That’s Measured, Not Claimed

MIX blends the processed signal against a latency-matched dry signal, so adjusting MIX to a partial wet/dry balance doesn’t introduce the phase cancellation a time-misaligned blend would cause between wet and dry paths. DRIVE saturates the processed effect signal specifically, not the original source passing through it, so pushing DRIVE adds harmonic content to the formant-shifted character rather than to the underlying performance itself.

At SHIFT 0, GENDER 0, and no TILT or FOCUS coloring applied, the processing gain equals exactly 1 and the output nulls against the input to better than 75 dB down — RTM Audio states this is a measured result, tested at the exact latency the plugin reports rather than requiring separate compensation to verify.

SOLO isolates the difference signal the processing itself is adding, rather than the wet or dry signal alone, so a producer can hear exactly what a given SHIFT, GENDER, or TILT setting contributes before deciding how far to push MIX.

TRIM’s output gain range is fixed at ±12 dB, so compensating for a level change larger than that from an extreme SHIFT or GENDER setting needs a separate gain stage after THROAT GOAT. Producers running subtle character moves get all the gain-matching they need from TRIM alone; producers pushing extreme identity changes that shift perceived loudness further may need an additional trim stage downstream.

No Pitch Control Inside This Plugin

Running a pitch-shifter before THROAT GOAT changes what note reaches the formant warp; running one after changes the note the listener hears — the plugin’s own output stays anchored to whatever pitch arrived at its input, unconditionally.

FAQs