Screen #43 of 60 Batch 8 desktop • 1440x900px

Desktop Whisper AI Audio Transcription & Subtitle Generator (WAV/MP3 -> SRT/VTT)

Specification & Codex Reference Prompt (Click to expand) (Speech AI studio compiling podcast/interview audio into timestamped subtitles (SRT, VTT, JSON) with speaker diarization.)

Screen Role: Speech AI studio compiling podcast/interview audio into timestamped subtitles (SRT, VTT, JSON) with speaker diarization.

Design Intent: A modern AI audio transcription and subtitle generator studio UI at 1440x900 resolution in bright Light Mode. Clean off-white background (#F8FAFC). Center stage: Audio waveform scrubber at the top with a blue playhead at 01:24. Below is a clean, editable 2-column subtitle transcript table: - Row 1: "[00:00:02.100 -> 00:00:05.400]" Speaker 1 badge: "Welcome back to the SaaS Architecture podcast." - Row 2: "[00:00:05.800 -> 00:00:09.200]" Speaker 2 badge: "Today we are discussing in-browser WebAssembly engines." Left control sidebar (320px wide, white container with 1px border): - AI Model: [ OpenAI Whisper Small (Local WASM) v ] - Target Format: [ SubRip Subtitles (.SRT) v ] - Language: [ Auto-Detect (English 99.4%) v ] - Punctuation & Capitalization: Toggle ON Top export button in solid indigo "Export Synchronized SRT & VTT Subtitles". Clean audio software UI, zero emojis.

Component Architecture: Django Cotton components (<c-file_converter.* />) with Tailwind tokens matching Labb.io design language.

Direct route: /file-converter/studio/whisper/ Template: file_converter/screens/screen_43.html

Live Django Implementation (desktop)

Isolate →

Whisper AI Audio Transcription Studio

Speech-to-Text → SRT / VTT / JSON with Speaker Diarization

Neural AI Model

OpenAI Whisper Small (Local WASM)

Target Format

SubRip Subtitles (.SRT)

Language Detection

Auto-Detect (English 99.4%)

Punctuation & Capitalization

podcast_ep42_master.wav Playhead: 01:24.300 / 08:45.000

Synchronized Transcript

[00:00:02.100 → 00:00:05.400] Speaker 1

"Welcome back to the SaaS Architecture podcast."

[00:00:05.800 → 00:00:09.200] Speaker 2

"Today we are discussing in-browser WebAssembly engines."

Original Codex Reference Mockup Full PNG