A sovereign voice AI appliance — 2U, on-prem. A single chassis ships the entire voice AI platform inside: real-time and recorded speech-to-text, language identification, and operational APIs — in your rack, under your control.

Why operators choose Auricus Voice

  • Sovereignty & control — audio and transcripts stay inside your network. Designed for finance, healthcare, public sector, defense, and any deployment where cross-border processing or third-party AI providers are a non-starter.
  • Predictable economics, end-to-end — single capital line item. No per-minute API meter, no egress, no rate-limit surprises. See the savings analysis for the matched-workload comparison.
  • Reliability — no dependency on external API uptime or internet jitter. Latency is bounded by your rack.
  • Integration — REST ingestion, async job queues, results by poll or webhook. Drops into existing ETL, CRM, and analytics stacks.
  • Operational visibility — Prometheus metrics + Grafana dashboards, structured audit logs, exportable SLO tracking.

Product family

Same model, same pipeline, same APIs across the family. Pick chassis density to match workload.

SKU Concurrent live calls Matched annual audio (M min/yr) Batch (files/min) End-to-end (files/hr) Peak accelerator power Target use case
Auricus Voice 8 12 2 25 60–75 80 W Mid-size contact center, departmental fleet
Auricus Voice 16 24 4 50 120–150 160 W Large enterprise contact center, regional carrier
Auricus Voice 32 48 8 100 240–300 320 W National contact center, telco / government scale

Concurrent live calls = sustained concurrent near-real-time conversations per appliance. Matched annual audio = realistic operational capacity per appliance under a typical mixed real-time + batch duty cycle. Batch (files/min) = sustained transcription throughput per appliance. files/hr (E2E) = end-to-end including ingest, language ID, decode, and delivery overhead. Peak accelerator power scales linearly across the family; add ~120 W chassis baseline for total appliance draw.

→ Public spec sheet: Specifications · platform deep dive: Platform.

Headline performance

  • Live and near–real-time — windowed streaming for contact centers, supervision, and compliance workflows. ~48–64 concurrent near–real-time streams per Auricus Voice 32 appliance before queue limits dominate.
  • Batch at scale — sustained ~100 files/min on Auricus Voice 32; end-to-end (ingest, language ID, delivery) ~240–300 files/hour under typical overhead.
  • Languages — ~99 transcription targets and 107 automatic language-ID classes.

Footprint

  • Auricus Voice 32 at full load: ~600 W typical appliance draw — vs comparable GPU-based racks measured in kilowatts.
  • 2U chassis — one rack-unit-class enclosure replaces multi-server GPU racks, cutting embodied materials, cooling load, and end-of-life e-waste.

Compliance posture

  • EU — AI Act 2 August 2026 ready — vendor-side primitives for Articles 9–17 (provider) and Article 26 (deployer) obligations under Regulation (EU) 2024/1689 where the deployer’s use case is high-risk; GDPR-aligned with on-prem retention, structured audit logs, and the architecture proposed in the EU Digital Omnibus.
  • US — sector-ready — primitives for HIPAA (BAA available), GLBA Safeguards Rule, PCI DSS 4.0.1, and CJIS; mappings to NIST CSF 2.0, NIST AI RMF 1.0, NIST SP 800-53 Rev. 5 and 800-171 Rev. 3 / CMMC 2.0 for federal and defense customers.
  • State-level — biometric-privacy aware (Illinois BIPA, Texas CUBI, Washington H.B. 1493), CCPA / CPRA sensitive-PI ready, and aligned with the Colorado AI Act (in force 1 Feb 2026).
  • On-prem only — no third-party processor in the inference path; no external egress required.
  • TLS in transit, bearer-token authentication on the REST API, structured audit logs with request IDs.

→ Full posture (EU + US): Compliance.

Get started

ask@auricus.com — pilots, pricing, and integration questions.