Public artifacts establish shared language. Private engagements address real risk. Each pathway below routes directly to its own inquiry process - please select the one that reflects where you are and what you need.
All inquiries are reviewed personally. Priority review is reserved for teams launching within 14-60 days.
Please select the pathway that reflects your situation. If you are unsure, the Voice AI Audit is the most common entry point - it is faster to scope, lower in commitment, and frequently becomes the foundation for a longer relationship.
A confidential, independent perceptual evaluation of your voice AI system - identifying tonal sycophancy, ambivalence collapse, and cross-modal dissonance before they surface in public deployment. Three tiers available from rapid triage to full adversarial red-team. Most teams discover at least one significant tonal risk they did not know existed.
Access to the TonalityPrint™ dataset - a human-verified perceptual reference corpus for stabilizing tonal inference in native audio-reasoning models. Structured for fine-tuning, perceptual benchmarking, sycophancy mitigation, ambivalence calibration, and cross-modal coherence evaluation. A foundational alignment asset - not a branded voiceover product.
For independent researchers, academic teams, and safety groups exploring perceptual alignment methodology, prosodic interpretability, or tonal trust signals in voice AI. The TonalityPrint™ dataset and Tonality as Attention™ framework are archived on Zenodo for peer review. Collaborative validation and multi-speaker extensions are actively sought.
For senior leaders navigating voice AI strategy at the organizational level - not a single audit or dataset, but an ongoing advisory relationship for principals building at the frontier of human-AI interaction. This is not a consulting retainer with deliverables. It is access to independent, unaffiliated strategic thinking on perceptual alignment, trust architecture, and responsible voice AI deployment at scale. One engagement considered at a time. Accepted by fit, not application volume.
Inquiries by introduction or direct application. Reviewed personally within 48 hours.
Every engagement - regardless of pathway - begins with a confidential scoping conversation. This is not a sales call. It is a diagnostic exchange to confirm whether the work is relevant, whether the fit is right, and what the appropriate next step looks like. If meaningful engagement is not possible, that will be shared with you directly and promptly.
Availability across all pathways is intentionally limited to maintain the depth and independence that make this work custom to you. Priority goes to organizations with clear, time-sensitive need. Engagements are vetted & accepted selectively, based on alignment fit, deployment context, and availability, not on first-come volume alone.
The following materials are released publicly to establish a shared framework around perceptual alignment in voice systems. They represent a small subset of the full body of work - and are not intended to replace direct engagement.
A conceptual framework examining how human vocal tonality functions as an attention-shaping and trust-modulating signal in voice AI systems. Formalizes a class of perceptual failures many teams experience intuitively but struggle to articulate or measure. Intended for: research leaders, product teams, and executives seeking clarity around felt experience, trust, and voice-mediated interaction.
View on Zenodo - DOI 10.5281/zenodo.17410581A contrast-structured, single-speaker prosody dataset designed for perceptual calibration, interpretability, and tonal alignment research. Annotates five functional tonal intents - Trust, Attention, Reciprocity, Empathy Resonance, Cognitive Energy - plus a systematically annotated ambivalence condition treated as a perceptual entropy feature rather than noise. Intended for: researchers exploring inference-time prosodic alignment, ambiguity-aware dialogue systems, and safety-critical voice agents that must audibly signal uncertainty.
View on Zenodo - DOI 10.5281/zenodo.17913895A concise executive-facing overview of why perceptual instability emerges in modern voice systems - and why fine-tuning alone does not reliably resolve it. Designed to support rapid internal alignment and decision-making under time pressure. Intended for: CEOs, VPs, and senior research or product leaders navigating voice AI tonal risk.
Download Executive Overview PDFPlease select the pathway above that reflects where you are - or read more about the research that underpins this work.
→ Read Research Origin ←