CueProof is a free web app that helps competitive dog handlers and trainers build collision-resistant verbal cue sets. It scores how likely any two cues in your set are to be confused by a dog under real training conditions — noise, distance, motion, and stress — and suggests alternatives to resolve the riskiest overlaps.

Dogs don't process speech the way humans do. They rely on acoustic edges — onset consonants, vowel shape, and syllable rhythm — rather than the full phonemic detail that human listeners use. Two cues that sound clearly distinct to you may share the same onset energy, vowel pattern, or syllable count, making them acoustically similar from your dog's perspective.

This is supported by research from Root-Gutteridge et al. (2019), who showed that dogs perceive and normalize formant-related vowel differences in human speech, and Andics et al. (2016), whose neural imaging work showed that dogs use distinct brain regions to process word meaning and intonation, relying heavily on acoustic features.

Every pair of cues in your set is compared using 10 phonetic similarity metrics, each weighted by its relevance to how dogs hear. These include edit distance, vowel similarity, consonant skeleton matching, n-gram overlap, onset class, syllable rhythm, and edge (prefix/suffix) matching.

On top of the base score, each pair runs through 7 degradation tests that simulate real-world conditions: final consonant drop, vowel compression under rapid speech, weak /h/ drop at distance, unstressed syllable loss, high-frequency consonant confusion in noise, formant shift under shouting, and reverb-induced stop masking in indoor halls.

If any degraded form of one cue matches a degraded form of the other, a booster penalty is applied. The final risk score determines severity: below 0.35 is clean (green), 0.35–0.65 is moderate risk (amber), and above 0.65 is high risk (red).

For full technical detail, see the How we score page.

CueProof currently supports five disciplines with full exercise templates:

• FCI Obedience — 9–10 exercises across 3 classes
• FCI IGP (Internationale Gebrauchshundeprüfung) — 19 exercises across 3 classes
• FCI IBGH (International Companion Dog Test) — 5–8 exercises across 3 classes
• FCI Agility — 18 exercises (Agility + Jumping tracks, 3 levels each)
• FCI Hoopers — 5 obstacle types, 3 FCI classes (H1–H3) + national H0

Each discipline includes per-class exercise filtering, so you only see the exercises relevant to your competition level.

CueProof supports three core languages with full template candidates and dictionary coverage: English, Slovene, and German.

Over a dozen additional languages are available with dictionary-backed suggestions, including Czech, Hungarian, Dutch, Finnish, Swedish, French, Italian, Spanish, Portuguese, Polish, Danish, Norwegian, Russian, and Japanese — plus regional variants like Belgian French and Belgian Dutch.

You can freely mix languages within a single cue set — common in competition, where handlers use their native language for some exercises and a foreign language for others to increase distinctiveness.

Yes. CueProof is free to use. You can create cue sets, run full scoring, see collision severity, read plain-language reasons for each collision, and get auto-suggested alternatives — all without paying.

Signing in with Google unlocks additional features: shareable links, PDF export, snapshot comparisons, and the minimal-change plan with a score improvement curve.

Auto-suggest uses a greedy optimization algorithm. Starting with an initial candidate for each exercise, it scores all pairs, identifies the exercise involved in the worst collision, tries every alternative candidate for that exercise, keeps the best swap, and repeats — up to 50 iterations or until no amber or red collisions remain.

Candidates come from the built-in dictionary (990+ entries across all supported languages) and discipline-specific template suggestions. In mixed-language mode, the optimizer draws from all language pools simultaneously.

The minimal change plan answers: "What is the fewest cue swaps I need to make to get the biggest score improvement?" It runs the same greedy algorithm as auto-suggest but captures the score after each swap, showing you a step-by-step trajectory.

You can lock exercises whose cues you don't want to change (because your dog already knows them reliably), and the optimizer will work around those constraints.

CueProof's scoring model draws on two research lines: confusability detection in speech technology, and dog speech perception research. Key studies include:

• Root-Gutteridge, Ratcliffe, Korzeniowska & Reby (2019) — dogs perceive and normalize vowel differences in human speech
• Andics, Gabor, Gacsi, Farago, Szabo & Miklosi (2016) — neural mechanisms for lexical processing in dogs
• Miller & Nicely (1955) — perceptual confusions among English consonants
• Summers, Pisoni, Bernacki, Pedlow & Stokes (1988) — effects of noise on speech production and perception
• Heffner (1983) — hearing thresholds in large and small dogs
• Liberman, Delattre, Cooper & Gerstman (1954) — the role of consonant-vowel transitions in stop consonant perception

The full list of citations with journal references is available on the How we score page.

Without signing in: your cue sets are stored locally in your browser (IndexedDB). Nothing is sent to a server.

After signing in with Google: your email, display name, and cue sets are stored in a Cloudflare D1 database in the EU. Session cookies are HttpOnly, Secure, and SameSite=Strict with a 1-hour expiry.

CueProof uses no advertising or tracking cookies. You can delete your account and all associated data at any time from the navigation menu. For full details, see the Privacy Policy.

All phonetic scoring runs in a Web Worker inside your browser — no server round-trips are needed for scoring. If you've loaded the app while online, anonymous cue sets (stored in IndexedDB) remain accessible. However, signing in, sharing links, and PDF export require a network connection.

The scoring model is a calibrated approximation grounded in peer-reviewed research, not a guarantee of error reduction. Dogs vary in hearing acuity, breed, age, and training history. Environmental factors (venue acoustics, handler voice, ambient noise) also affect real-world discriminability.

CueProof's value is in identifying pairs that are phonetically closer than you might realize and suggesting alternatives that increase acoustic contrast. Think of it as a second opinion on your cue set, not a definitive verdict.

Yes. Signed-in users can generate a shareable link that gives read-only access to the scored cue set — including the score ring, collision cards, and severity breakdown. The link works without the recipient needing to create an account.

You can revoke any shared link at any time from your dashboard or the report page. Once revoked, the link returns a 404 — it cannot be reactivated.