Voice + Haptic + Simplified
Multi-modality access with continuous fairness audit — every cohort counts, every cohort tracked.
Sesim Inclusive Access (Layer 4) reaches every voice — visual, hearing, motor, cognitive and elderly cohorts — with continuous bias detection (Patent claim 53) across accent, age, gender and language-switch deltas; production halts for any cohort that breaks the threshold. Multi-jurisdiction compliance built-in: KVKK m.6 + GDPR Art. 9 + EU Accessibility Act 2019/882 + EN 301 549 + ADA Title II/III + AODA + BDDK §11 — one SDK. Combined with Layer 1 (coercion detection) and Layer 2 (voice identity), one voice intent yields four signals: who + under coercion? + what action + accessibility need.
Live Layer 4 deployments require a partner-led pilot with a public agency, NGO, or regulated bank. To request access for an e-government, kiosk, ATM, healthcare MIS, or accessibility audit pilot, contact us.
Request pilot access →A typical 4-modality access turn — illustrated only. No real audio is captured on this page; live access requires a partner-led pilot.
- 1Modality detect
screen-reader: yes · haptic: yes · mic-quality: HD · accent: TR-Karadeniz - 2User says
"Pasaport başvurusu yapmak istiyorum" - 3ASR + intent
{ action: NAVIGATE, target: passport_application, confidence: 0.91 } - 4Bias check
cohort: accent_karadeniz · delta: 0.04 · status: HEALTHY - 5L1+L2 cross-layer
L1 stress 0.12 · L2 voiceprint 0.94 → identity confirmed - 6Modality response
voice: "Pasaport başvurusu açılıyor" · haptic: short-pulse · visual: focus on form - 7Audit + log
voice intent + bias score + WCAG conformance hash · 7-year retention
How 4-modality access works
- Voice guidance: TTS responses adapt to user speed (slow / normal / fast) and persona (male / female / neutral).
- Haptic confirmation: short / long / pattern vibrations replace audio cues for hearing-impaired users.
- Simplified-command vocabulary: short verb + noun structure; every action mapped to a clear word (Patent claim 16).
- Multi-language code-switching: TR/EN in a single utterance handled by language-segment ASR (Patent claim 49).
- Bias detection (Patent claim 53): accent / age / gender / language-switch deltas continuously monitored. >25% delta = production halts for that cohort.
- WCAG 2.2 AA SDK conformance: 47 success criteria automatic, 18 partner-environment guidance.
Why public-grade not consumer-grade
- Consumer accessibility tools (NVDA, JAWS, VoiceOver) are screen readers — one-way (screen → speech). Sesim L4 is two-way (speech ↔ speech + haptic).
- No commercial competitor offers integrated bias-detection guardrail with public reporting.
- EU Accessibility Act (2019/882), EN 301 549, ADA Title II/III, AODA, BDDK §11 — all aligned in one SDK.
- KVKK m.6 / GDPR Art. 9 explicit-consent flow built in: disability status = sensitive data, 7-year audit retention.
- Patent moat (Türkpatent claims 16, 25, 27, 49, 53, 109) — category-protective IP for "Accessibility-as-a-Service".
- Cross-layer with L1: elderly coercion detection (Patent claim 47 silent-alarm) — protects vulnerable users from family / caregiver pressure.
KPI gates (pilot)
- Task-completion rate +30% vs. baseline in target segment.
- Time-to-completion -30%; error rate -50% vs. baseline.
- WCAG 2.2 AA external audit PASS (TÜBİTAK BİLGEM or W3C-aligned firm).
- Bias delta ≤ 10% across accent / age / gender / language-switch cohorts.
- Latency ≤ 800ms p95 (voice command → haptic confirmation).
- NPS ≥ +40 in target segment (≥ +50 for elderly cohort).