PROGRAM BRIEFING DOCUMENT

WHISPER

PROGRAM REF: HYK-003 · CODENAME: WHISPER

CLASSIFICATION

AI SYSTEMS / SPEECH-TO-TEXT

STATUS

ACTIVE

TECH STACK

Next.js · React 19 · TypeScript · Drizzle ORM · GSAP · Tailwind CSS 4

INFRASTRUCTURE

OpenAI Speech-to-Text API · PostgreSQL · Auth.js 5 · Dokku

MISSION BRIEF

Audio transcription platform powered by OpenAI's Speech-to-Text models. Operators upload audio files and receive transcripts streamed in real time over Server-Sent Events. Three models — GPT-4o Transcribe, GPT-4o Mini, and Whisper-1 — run side-by-side for comparison, with export to plain text, JSON, SRT, and VTT. Oversized files are split into overlapping chunks and stitched back into a single unbroken transcript. GitHub-authenticated access; every transcript archived to PostgreSQL.

CAPABILITIES

Multi-model transcription via OpenAI (GPT-4o Transcribe, GPT-4o Mini, Whisper-1)
Real-time streamed results over Server-Sent Events
Side-by-side model comparison on identical audio
Subtitle and transcript export — TEXT / JSON / SRT / VTT
Chunked transcription of large audio files with overlap stitching
Transcript history archived in PostgreSQL behind GitHub OAuth

SOURCE: CLASSIFIED ACCESS DEPLOYMENT →

← RETURN TO PROGRAM REGISTRY