The intelligence layer for real-time communication.

ODIN Cortex — The Intelligence Layer

Turn raw audio traffic into actionable insights and automations. Real-time transcription, hate-speech detection, AI summaries, serverless functions, and event webhooks — all natively integrated into the ODIN stack.

Start Building Free Read the Docs

Made with ❤️ by 4Players in Germany 🇩🇪

Trusted by studios and enterprises building intelligent communication.

More Than Transcription. An Intelligence Platform.

Cortex processes your audio streams in real-time. Transcribe voice chat, detect toxic speech, generate meeting summaries, trigger webhooks, and run custom serverless functions.

Transcription

Real-Time Speech-to-Text

Whisper-powered transcription runs directly on your ODIN Voice streams. No external pipeline needed. Results stream in real-time via webhooks or REST API.

40+ languages supported
Speaker diarization
Automatic punctuation
Streaming & batch modes

Safety

Hate-Speech Detection

Voice + NLP + Context = accurate toxicity scoring. Configurable thresholds, automatic sanctions, and full audit trails for compliance.

Real-time scoring
Configurable thresholds
Auto-sanctions & muting
Full audit trails

summaries

Meeting Summaries

Auto-generate structured summaries from voice sessions. Action items, key decisions, and follow-ups.

Native

integration

Voice + Cortex

Cortex processes ODIN Voice audio directly. No second stream, no external service, no data transfer.

Functions

Serverless Functions

Write custom JavaScript/TypeScript code that runs on Fleet infrastructure. Event-based triggers, HTTP endpoints, and a Monaco editor in the dashboard.

JS/TS runtime on Fleet
Event-driven triggers
HTTP endpoints
Browser-based editor

Events

Event Pipeline & Webhooks

Every Cortex action emits events. Subscribe via webhooks with retry logic, or consume via REST API. Build custom workflows without polling.

Webhook retry with backoff
Server-sent events
REST API for all operations
Custom event filters

Knowledge

Semantic Search

Vector-based semantic search across all your transcriptions and sessions. Ask questions in natural language and get contextual answers from your communication history.

pgvector similarity search
Session archive with full text
Natural language queries
Contextual AI answers

€0.006

per minute

Simple Pricing

All features measured in Cortex Minutes. One metric, one price.

Roadmap

Sentiment Analysis

Real-time emotional tone analysis of conversations. Track player sentiment and customer satisfaction.

Roadmap

Real-Time Translation

Break language barriers. Automatic translation of transcribed speech for multilingual voice chat. 40+ language pairs with subtitle overlay option.

Cortex vs. Building It Yourself

DIY Approach

Building audio intelligence from scratch requires stitching together multiple services, managing infrastructure, and months of integration work.

Route audio to external transcription API
Build custom hate-speech pipeline
Manage your own serverless infrastructure
Wire up webhooks and events manually
Separate billing for each service
Months of integration work

With ODIN Cortex

Cortex is natively integrated into the ODIN stack. Enable features in your dashboard and start using them immediately.

Transcription runs natively on Voice streams
Built-in toxicity detection with configurable rules
Serverless functions on Fleet infrastructure
Event pipeline with webhooks out of the box
One bill, measured in Cortex Minutes
Enable in your dashboard, start using today

Made for Developers

Enable Cortex features with a few lines of code. Real-time transcription, voice moderation, and custom serverless functions — all through one unified API.

TypeScript

Real-Time Transcription

Enable Cortex transcription and get streaming results in real-time.

cortex-transcription.ts

import { OdinClient } from "@4players/odin";

const room = await OdinClient.initRoom(token);
await room.join();

// Enable Cortex real-time transcription
await room.enableCortex({
  transcription: { enabled: true, language: "en" },
  moderation: { enabled: false },
});

// Transcription results stream in automatically
room.addEventListener("TranscriptionResult", (ev) => {
  const { peerId, text, isFinal } = ev;
  if (isFinal) {
    console.log(`[${peerId}]: ${text}`);
    appendToChat(peerId, text);
  }
});

TypeScript

Voice Moderation

Add toxicity detection with automatic sanctions in a few lines.

cortex-moderation.ts

import { OdinClient } from "@4players/odin";

const room = await OdinClient.initRoom(token);
await room.join();

// Enable voice moderation with toxicity scoring
await room.enableCortex({
  transcription: { enabled: true },
  moderation: {
    enabled: true,
    threshold: 0.7,
    actions: ["flag", "mute"],
  },
});

// React to moderation events
room.addEventListener("ModerationEvent", (ev) => {
  const { peerId, category, score } = ev;
  console.log(`Toxicity: ${category} (${score})`);
  if (score > 0.9) room.mutePeer(peerId, 30);
});

TypeScript · Serverless

Cortex Functions

Custom backend logic triggered by Cortex events, running on Fleet.

cortex-function.ts

// Cortex Serverless Function — runs on Fleet
export async function onTranscription(
  event: CortexEvent
): Promise<void> {
  const { roomId, peerId, text } = event;

  // Store the transcription
  await db.transcriptions.insert({
    roomId,
    peerId,
    text,
    timestamp: Date.now(),
  });

  // Forward to external webhook
  await fetch(WEBHOOK_URL, {
    method: "POST",
    body: JSON.stringify({ roomId, peerId, text }),
  });
}

Use Cases for Cortex

How teams are using Cortex to add intelligence to their communication.

Gaming

Game Studios

Anti-toxicity compliance for voice chat. Automatic transcription for tournament replays. Custom serverless functions for match events and leaderboards.

Real-time voice moderation
Tournament replay transcription
Custom event-driven logic
Player behavior analytics

Enterprise XR & Training

Meeting transcription and summaries for VR training sessions. Session archives with semantic search. GDPR-compliant processing.

SaaS

Communication Apps

White-label transcription for conferencing products. Custom moderation rules per tenant. Serverless workflow automation.

Part of the ODIN Platform

Cortex is the intelligence layer of the ODIN Platform. It works seamlessly with Voice, Fleet, and Rooms.

ODIN Voice

Cortex processes Voice audio streams natively. No external routing or data transfer needed.

Explore Voice

ODIN Fleet

Serverless Functions run on Fleet infrastructure. Deploy custom backend logic without managing servers.

Explore Fleet

ODIN Rooms

Meeting transcription and summaries power the next generation of Rooms conferencing features.

Explore Rooms

Cortex Pricing

All features measured in one simple metric: Cortex Minutes. Included in every platform tier.

Free

€0

/ month

60 Cortex Minutes
~1 hour of transcription.
1 Serverless Function
Basic automation.
Webhooks
Free with all tiers.
Community Support
Discord & docs.

Start Free

No credit card required

Starter

Pay-as-you-go

Scale

unlimited

€0.006 / Cortex Minute
Volume pricing.
Unlimited Functions
No limits.
Full Feature Access
Everything included.
Dedicated Support
Enterprise-ready.

View Full Pricing

Custom contracts available

Cortex Minutes are included in platform tiers • 1 min audio = 1 Cortex Min • Toxicity check = 0.1 min • Webhooks always free

Add Intelligence to Your Communication

Start with 60 free Cortex Minutes. No credit card, no commitments. Enable transcription, moderation, and custom functions in your ODIN dashboard today.

Start Building Free Read Documentation Talk to Sales