Baruch — Broadcast TV Live Captioning

Technical Overview

This technical overview details Baruch’s live captioning solution for broadcast television, covering system architecture, ASR integration, workflow compatibility, formatting logic, and deployment models. It outlines support for legacy and modern subtitle delivery standards used in European and international broadcast environments, including Teletext (via Newfor) and TTML/EBU-TT-D.

Broadcast-Grade Live Captioning Engine

Baruch is engineered for on-air use in live broadcast environments where latency, accuracy, and reliability are critical. The captioning engine delivers 1–2 seconds end-to-end latency, keeping live subtitles for linear TV in sync with live speech. Subtitle delivery supports European broadcast standards, including Teletext (via Newfor) and TTML/EBU-TT-D.

The solution is designed for 24/7 playout operations, supporting continuous channel output and dynamic programme changes. Baruch integrates with both SDI and SMPTE 2110/IP infrastructures, allowing deployment in legacy or fully IP facilities.

ASR-Agnostic Speech-to-Text Layer

Baruch operates independently of any single ASR vendor and supports integration with multiple broadcast-grade ASR engines, ensuring flexibility in accuracy, language coverage, and commercial choice. The ASR component can be deployed in the cloud, on-prem, or in a hybrid configuration, enabling broadcasters to align captioning architecture with internal security, latency, and data-handling policies.

Hybrid deployments allow Baruch to run on-prem for full control of ingest and output, while ASR operates in the cloud for scalability. For high-security or air-gapped environments, both the Baruch application and the ASR engine can be deployed fully on-premises. This model supports low-latency processing of live subtitles for linear TV and eliminates external data exposure, meeting strict compliance requirements.

Deployment & Architecture Options

The Baruch application is installed on-premises within the broadcaster’s environment, ensuring direct control of ingest, processing, latency, and output routing for live subtitles for linear TV. Broadcasters can run the ASR engine in the cloud or on-prem, depending on operational, security, and regulatory requirements.

A typical hybrid architecture runs Baruch on-prem for deterministic performance and local resilience, with ASR in the cloud for scalability and continuous model updates. For secure or air-gapped environments, full on-prem deployments of both Baruch and ASR are supported. This architecture enables deterministic latency, internal data retention, and integration with broadcaster IT and MCR security policies.

Subtitle Formatting, Filtering & Editorial Controls

Baruch provides a broadcast-specific subtitle formatting layer that adapts raw ASR output to on-air captioning standards. Broadcasters can configure segmentation, timing, reading-speed limits, and character-per-line rules to ensure live subtitles for linear TV remain readable within broadcast editorial guidelines. Multiple rendering modes are available, including pop-on, progressive (rolling-line), and multi-line layouts, with per-channel configuration.

The system includes custom dictionaries with update capability for names, terminology, and event-specific vocabulary. Filtering algorithms improve the readability of live subtitles for linear TV by removing filler words, repeated phrases, and hesitations. Profanity filtering is included for compliance. Automatic casing, punctuation, and sentence-boundary detection produces cleaner subtitles without manual intervention. Broadcasters can apply language-specific profiles to tailor formatting rules per channel or territory.

Baruch includes an operational monitoring interface for channel monitoring and status visibility, and supports updating dictionaries and channel profiles as needed without disrupting live output.

Integration Into Broadcast Workflows

Baruch integrates natively into live production and playout environments and is compatible with industry-standard broadcast captioning and graphics systems, including Pixel Power, BroadStream, Evertz, and Newfor. Baruch ingests live programme audio from SDI or SMPTE 2110-based environments for real-time transcription of live subtitles for linear TV, and delivers subtitle data to downstream encoder systems where insertion and playout are handled.

For linear TV distribution, Baruch supports Teletext subtitles (via Newfor) for compatibility with traditional broadcast delivery across European markets. Baruch also supports TTML/EBU-TT-D for modern and future-ready subtitle workflows. Output can be passed to existing caption encoder infrastructure for insertion into the outgoing signal chain, ensuring full compatibility with current broadcast systems.

The integration design enables fully automated or human-assisted operation, supporting workflows with or without existing respeaker involvement for live subtitles for linear TV.