Transcription Platform

NexusScribe

AI-Powered Real-Time Transcription & Meeting Intelligence Platform

A self-hosted, enterprise-grade transcription platform that transforms meetings into actionable intelligence. Built for organizations requiring on-premises data sovereignty, it delivers real-time speech-to-text with speaker diarization, powered by local AI inference on CPU, CUDA GPU, or Hailo NPU.

For more information, contact us at devops@automatanexus.com

Download NexusScribe v1.0.0

Pre-built binaries or build from source

BSL-1.1 License

Free Use: Internal use, development, testing, and non-profit organizations

Commercial: Providing commercial transcription services requires a license agreement

Future: Converts to Apache-2.0 open source on January 26, 2030

Complete Meeting Intelligence

From live transcription to AI-powered insights, NexusScribe handles the entire workflow

Real-Time Transcription

Sub-200ms latency with Whisper models

Speaker Diarization

MFCC-based voice separation & identification

Meeting Intelligence

Summaries, action items, sentiment analysis

Live Collaboration

Operational transformation for real-time editing

Video Integration

Teams, Zoom, Google Meet, LiveKit support

Mobile App

Cross-platform iOS/Android with Expo SDK 54

Key Features

Multi-Backend Inference

CPU (64GB+ RAM), CUDA GPU, and Hailo NPU (8/10-H) compute backends with automatic language detection for 99 languages.

Enterprise Security

Role-based access control, multi-factor authentication (TOTP), JWT with refresh rotation, SOC 2 and HIPAA compliant audit logging.

AI Studio Tools

Edit for clarity, professional tone adjustment, filler word removal, auto-chapters, action item extraction, and speaker identification.

Export Formats

PDF, DOCX, JSON, SRT, and VTT exports with timestamp-linked commenting and version history with full audit trail.

Meeting Bot Integration

Automated bots join Microsoft Teams, Zoom, and Google Meet to capture audio and provide real-time transcription.

Aegis-DB Backend

Multi-paradigm database for users, meetings, transcripts (relational), audio metrics (time series), and topic relationships (graph).

Mobile App (React Native / Expo)

Cross-platform iOS and Android support with real-time transcription

Dashboard

Quick access to recording, meetings, AI Studio

Recording

Real-time audio level visualization

AI Studio

Waveform playback with editing tools

External Meetings

Connect bots to Zoom/Teams/Meet

Offline Mode

Queue actions, sync when online

Share & Export

TXT, JSON, PDF, DOCX formats

Data Flow

Audio Input → Preprocessing (16kHz mono, VAD) → Inference Engine →
Speaker Diarization → WebSocket Broadcast → Aegis-DB Storage →
Collaborative Editor → Export (PDF/DOCX/JSON/SRT)

Technical Specifications

BackendRust (Axum) + 12 Crates
FrontendNext.js Web + React Native Mobile
DatabaseAegis-DB (4 paradigms)
InferenceWhisper.cpp (CPU/CUDA/Hailo)
Process ManagerPM2
Modelstiny, base, small, medium, large-v2/v3
Latency<200ms (Hailo), <100ms (CUDA)
Languages99 supported

Performance Benchmarks

<200ms
RPi 5 + Hailo 10-H
15 concurrent streams
<100ms
x86 + CUDA RTX 3080
50 concurrent streams
2-3%
WER (Studio Audio)
large-v3 model

Ready for Enterprise Transcription?

Contact our team to learn how NexusScribe can transform your meetings into actionable intelligence with on-premises data sovereignty.

devops@automatanexus.com