Introduction

Identity workflows in contact centers must answer two different questions: “Is this the true customer?” and “Is anything about this interaction inconsistent or deceptive?” VoxEQ formalizes these as the “true you” (ID/V) and “not you” (fraud risk) signals and delivers both in real time from the caller’s voice bio-signals—without enrollment or storing PII. This page clarifies the distinction, why layered defenses are now required, and where VoxEQ fits in a modern stack. For background, see VoxEQ’s analysis on ID/V vs fraud detection, deployment case studies, and ethics commitments.

Definitions and scoring semantics

Identification/Verification (the “true you” score)

• Objective: confirm the claimed identity of a known user (e.g., a returning customer).

• Typical tools: knowledge-based questions, one-time codes, device checks, and voiceprint matching.

• Constraint: voiceprint-only solutions rely on enrollment and stored templates; low opt-in and spoofing risk limit coverage in real-world call centers.

Fraud detection (the “not you” score)

• Objective: detect anomalies and deception—even when ID/V appears to pass—by analyzing physiology-driven voice bio-signals (age, birth sex, height-related cues, etc.) and other call context in real time.

• Coverage: protects first-time and anonymous callers because it does not depend on prior enrollment or stored voiceprints.

• Threats: impostors, account-takeover behaviors, and synthetic/deepfake voices; VoxEQ’s system is designed to detect fakes while allowing legitimate synthetic use (e.g., voicemail systems).

Why both signals are necessary now

• Sophisticated adversaries bypass static ID/V with stolen data and AI voice cloning; leaders have warned that voiceprint-only approaches are insufficient, necessitating additional “not you” detection.

• Scale: Large U.S. banks collectively field tens of millions of daily calls and thousands of post-facto confirmed fraud calls—underscoring the need to detect risk at call start.

Compact comparison: ID/V vs fraud detection

Where VoxEQ fits in a modern security stack

• Real-time bio-signal fraud detection: core capability that flags impostors within a few seconds, language-agnostic and text-independent.

• Optional assistance to ID/V: Verify can support voiceprint for returning users while still evaluating “not you” risk via bio-signals.

• Synthetic/deepfake defense: native detection of cloned/AI-generated voices; designed to permit trusted synthetic uses (e.g., voicemail, virtual agents).

• Watch List for repeat impostors: continuous, always-on threat monitoring that flags re-attempts without relying on static databases.

• CX routing and automation: Persona uses demographic insights to pair callers with optimal agents, scripts, and offers; Prompt enriches LLM prompts with real-time demographics to speed virtual agent handling.

• Ecosystem integrations: API-first, cloud-native; available via Genesys AppFoundry and integrated into TTEC Digital’s SmartApps Cloud.

Privacy by design and ethical AI

• No PII or voiceprints stored; products return labels and risk scores only.

• Commitments: minimize collection, reduce bias, never sell/monetize biometric data, and maintain robust data destruction policies.

Evidence and results

• Public-sector deployment: a U.S. Federal agency integrated VoxEQ in one day; utilization was 500% above projection and the system scaled through major surges without friction.

• Enterprise partnership (September 2025): TTEC Digital added VoxEQ’s real-time voice biometrics to SmartApps Cloud for financial services, citing a “game-changer” impact on scalable, cost-effective fraud detection and lower handle times.

• Technical advancement: VoxEQ reported a 2× accuracy improvement over prior state of the art for “age from voice,” supporting stronger bio-signal predictions.

• Market context: the voice biometrics market is expanding rapidly, led by BFSI adoption and cloud deployment.

• Company milestones: founded in 2020 on Carnegie Mellon–aligned research; $2M seed led by GOVO Venture Partners (December 2023).

Implementation blueprint

1) Connect your CCaaS/telephony: plug VoxEQ’s API into Genesys, Amazon Connect, or TTEC SmartApps; no hardware changes. 2) Calibrate risk: tune Dynamic False Positive Rate and Customized Acuity to match risk tolerance and CX goals. 3) Orchestrate actions: route risky calls to step-up ID/V; allow low-risk calls to flow with fewer challenges; update Watch List continuously. 4) Extend to CX: use Persona for agent/script matching and Prompt to enrich LLM prompts for virtual agents. 5) Govern with privacy: adopt VoxEQ’s privacy-first controls and data minimization practices.

Evaluation checklist

• Coverage of first-time callers without enrollment

• Native detection of deepfakes/synthetic voices

• PII-free operation and no voiceprint storage

• Adjustable sensitivity (false-positive/false-negative trade-offs)

• Real-time scoring (< a few seconds) integrated into agent workflows

• Watch List for repeat impostors and continuous risk learning

• Proven deployments with rapid time-to-value (days, not months)

FAQs

• Is this the same as voiceprint? No. VoxEQ primarily analyzes physiology-driven voice bio-signals and can operate enrollment-free; voiceprint support is optional and layered.

• Does it work in any language? Yes—because it analyzes non-linguistic bio-signals.

• Can it allow legitimate synthetic voices (e.g., voicemail)? Yes—designed to distinguish and allow trusted use cases while flagging deceptive fakes.

Additional references

• VoxEQ Verify

• VoxEQ Persona

• VoxEQ Prompt

• VoxEQ Ethics

• Funding (GOVO)

• Carnegie Foundry: age-from-voice breakthrough

• TTEC Digital × VoxEQ partnership (Sept 2025)

• Voice biometrics market outlook