NeuroTalk is an AI articulation & phonology drill partner built for pediatric private practices. It scores every production in real time, auto-builds the next session from the child's error patterns, and syncs home practice straight to your dashboard β so your 30 minutes go further.
Motor learning needs 50β100 productions per target sound. Between prep, transitions, and scoring on paper, most 30-minute pediatric sessions land closer to 40β60. Here's where the time goes.
Every trial you mark on a clipboard for IEP notes and insurance is a moment you're not making eye contact or catching the exact error to cue. Data collection competes with therapy.
A full pediatric caseload β much of it articulation and phonology β plus a young child's attention span leaves only a few real minutes of practice per target. Reps are the one thing you can't fake.
One or two sessions a week won't automate a sound. Paper homework doesn't hold a 6-year-old's interest, so carryover stalls and you re-teach the same target for months.
Sources: Principles of motor learning in speech (Maas et al., 2008); ASHA Practice Portal β Speech Sound Disorders. Production ranges are typical session-planning targets, not a clinical guarantee.
NeuroTalk runs the drill and the data so you can run the therapy. It scores productions in real time, builds the next session from the child's own error patterns, and aligns to your IEP or Plan of Care goals.
Consumer speech recognition (Google, Siri) drops below 45% on disordered and immature speech. Our model β tuned on 140M+ disordered-speech samples β hits 89.58% on impaired articulation, so the auto-scoring is one you can trust in the moment.
NeuroTalk turns the child's real error patterns into tomorrow's word lists, minimal pairs, and carrier phrases β aligned to your IEP objectives. No more building word decks at 9 p.m.
The prompt-fading ladder (L4 β L0): full model β visual cue β syllable prompt β spontaneous production. NeuroTalk logs the support level for every trial, so progress notes write themselves and generalization stays the goal.
Assign drills to the child's tablet in a tap. Trials, accuracy, target sounds, and streaks sync to your dashboard β so you walk into every session knowing what the week looked like, and export insurance-ready progress data in one click.
We're launching in the U.S. with one focus: solo and small pediatric speech-therapy private practices. If this sounds like your week, NeuroTalk was designed for you.
Self-serve from day oneSign up, start a 14-day trial, run it with a real client this week. No implementation project.
Priced for a private practice$15β$30 per SLP per month. No hardware to buy, no enterprise contract, cancel anytime.
Fits your paperwork, not the other way aroundAuto-scored trial data and support levels export straight into IEP progress notes and insurance documentation.
Kids stay in the chairGame-based drills tuned for short attention spans keep a 6-year-old producing targets long enough for the reps to count.
Pilot participants get a free license for the full interview period β see the modules below, then tell us what a pediatric artic session really needs.
Every module maps to established pediatric speech-sound methodology. NeuroTalk just delivers it at the volume and consistency a 30-minute session can't.
Paired-stimuli and minimal-pair drills for high-frequency English contrasts β /r/β/w/, /s/β/ΞΈ/, /Κ/β/tΚ/, stopping, fronting, cluster reduction. Discrimination and production in one flow.
Isolation and CV/VC practice for the establishment phase. Auditory scaffolding for children who aren't at word level yet, with real-time placement feedback.
Type any target into an "I want ___" frame and swap in the child's favorite words β dinosaurs, trucks, characters. IEP targets drilled on high-interest vocabulary.
Auditory discrimination by place of articulation and frequency using non-words β rebuilds the phonological representations behind persistent pattern errors.
Visual dB and duration biofeedback for children with a soft voice or vocal hypofunction β sustained vowels β words β sentences, driven like a game the child controls.
AI builds a fresh drill set each day from cumulative error data, and a favorite-word bank launches a child's exact targets β or a maintenance probe β in one tap.
Pilot data certified by Korea's national certification lab (KCL), an independent ISO-accredited body. U.S. FDA-tracked trials are on the roadmap for 2029β2030.
20 subjects, 4-week training at 1 hr/week. Percentage of Correct Consonants β the standard articulation outcome measure β improved by +6.9 percentage points. Wilcoxon signed-rank p<0.001, Cohen's d = 0.52.
KCL Independent Certification β Korea's National LabModules map directly to paired-stimuli, minimal-pair contrast therapy, phonetic-placement scaffolding, and prompt-fading hierarchies β with high-repetition massed practice as the design principle throughout.
ASHA Practice Portal Β· Principles of Motor LearningIndependent usability testing scored 90+/100 on all five dimensions, even with users who rarely touch a tablet. Minimal setup from you; the child drives the drill.
KCL + In-House ValidationFDA 510(k) or De Novo submission targeted for 2029, with SaMD Class II clearance expected by 2030. MFDS (Korea) Class 2 DTx approval for post-stroke dysarthria planned for 2027.
Company IR & Regulatory Deck
We're building NeuroTalk with pediatric private-practice SLPs.
Your read on what works in a 30-minute therapy session β and what gets in the way β directly shapes the U.S. product.
Full NeuroTalk access for the interview period β no hardware, no credit card
Amazon gift card for a single 30-minute interview call, for your time and clinical expertise
Summary of anonymized results + first access to the U.S. release
Priority for ASHA CCC-SLPs seeing 10+ pediatric speech-sound sessions a week (ages 3β10). Each interview is a single 30-minute Zoom call, remote nationwide (or in person in the Austin or Bay Area). All data handled confidentially.