The Challenge

Traditional subtitling can take hours to process a single file, and a live broadcast cannot wait in a transcription queue. Widescreen AI needed one platform that could grind through large media volumes in batch and deliver translated captions on air at the same time, without the compromise usually forced between the two.

The Engineering Strategy

HashOne Global built the complete AION platform: the AI orchestration layer, transcription and translation pipelines, media processing, the SaaS application, billing, dashboards and the real-time broadcast infrastructure.

The core decision was to put eight ASR and translation providers behind unified interfaces. AION can select, switch and scale AI services without touching the core system, which matters in a field where the best provider changes every few months.

AION Live runs a dedicated real-time pipeline: SRT in, streaming transcription with provider failover, in-flight translation against project-specific glossaries, SRT out. Target caption latency under six seconds. Automated recovery, live monitoring and diagnostics run underneath, because a caption pipeline that needs a human to restart it is not a broadcast product.

The Result

• 8 ASR engines behind one interface
• 100+ languages supported
• Under 6s target live caption latency

Now powering Widescreen AI's SRT and Live offerings in daily production use, running thousands of transcription and translation jobs alongside live multilingual broadcast workflows.

Technology

• FastAPI
• PostgreSQL
• Celery
• Redis
• FFmpeg
• SRT
• Next.js
• React
• Stripe
• S3
• Multi-provider ASR and translation