Sense platform

The technical spine of real-time media intelligence

Universal Media Taps connect any source to the Senses. Capabilities run continuously across audio, video, and metadata. Deployments flex from on-prem to cloud without a fork in your stack.

Universal Media TapsCapabilitiesDeploymentsIntegrations

Universal Media Taps

One ingress layer for every source

A tap is a clean, AI-ready copy of your live stream, peeled off ingest so Senses can analyze without touching the viewer pipeline. One ingest, two jobs: watchable media for humans, analyzable media for AI.

  • Streaming: MPEG-TS, SRT, HLS, RTMP
  • File ingest: MP4, MXF OP-1a, MOV, broadcast-grade containers
  • API push: REST ingest endpoint for transcode output
  • Coming soon: SMPTE ST-2110, JPEG-XS

Live stream

One ingest

A single live feed comes in. No second ingest, no extra cables.

  • MPEG-TS
  • SRT
  • HLS
  • RTMP

Coming soon

  • SMPTE ST-2110
  • JPEG-XS

Main pipe

Playback keeps its own pipe

Operators keep watching. Preview and on-air stay independent. AI never steals the main feed.

Taps

AI-ready copies, on by need

Turning on a Sense is what creates its tap. Audio-only AI means an audio tap and nothing more, so unused plumbing costs nothing.

  • AudioClassify · Transcribe
  • VideoFrames for vision models
  • MetadataCaptions · SCTE-35

Senses emit findings as events on the timeline, never the raw media.

What Sense sees across audio, video, and metadata once taps are live.

Audio
Video
Metadata

Capabilities

What Sense detects, 24/7

Six detection domains run in parallel across every tap, every feed, every minute of the day.

Loudness & CALM compliance

Loudness & CALM compliance

Continuous ITU-R BS.1770, CALM Act, and EBU R128 monitoring per segment, break, and program — with timestamped evidence clips.

Closed caption QC

Closed caption QC

FCC Part 79 presence, timing, and accuracy checks across all tracks — for air and catch-up streams.

Audio content classification

Audio content classification

Speech, music, silence, and ambient detection per frame. Feeds loudness, rights, and language workflows.

SCTE-35 cue detection

Splice insert, splice null, and blackout trigger detection with insertion-rate anomaly alerting.

Real-time alerting

Real-time alerting

Configurable thresholds push to the operator surface, PagerDuty, Slack, and webhook endpoints — sub-second from detection.

Video QC

Video QC

Black frame, freeze frame, and signal-dropout detection with frame-accurate timestamps and operator clips.

Deployments

On-prem unmetered, cloud elastic

On-premises

Run inside your NOC. Fixed annual license with no per-stream overages, no cloud egress, and no data-sovereignty concerns.

Unmetered
  • 24/7 unmetered processing
  • Air-gapped or private-network deployment
  • Hardware appliance or VM on your own infrastructure
  • Predictable annual pricing
Get a quote

Cloud

Spin up capacity for events, overflows, and cloud-native pipelines — burst for the tentpoles, scale down between them.

Elastic
  • AWS, Azure, and GCP regions
  • Auto-scale to thousands of concurrent streams
  • Burst capacity for live events and tentpoles
  • Flat per-track rate — PAYG for one-off events
See cloud pricing

Integrations

Fits your existing stack

Pre-built connectors cover the broadcast infrastructure you already run. No rip-and-replace required.

  • AWS Elemental

    MediaLive and MediaConvert output taps

  • Harmonic

    ViBE and Spectrum encoder output

  • Grass Valley

    AMPP and Morpheus automation hooks

  • Telestream

    Vantage transcoding pipeline

  • Imagine Communications

    Versio and Selenio playout output

  • Webhook / REST

    Any pipeline with an HTTP output endpoint

Ready to instrument your pipeline?

Create an account, size your tracks, and subscribe. We provision on-prem licenses after checkout — or book a technical walkthrough with engineering.