Deepgram Nova-3 + Gemini Dual-Pass Audio/Video Caption Studio

Deepgram API Key (Nova-3 STT Baseline)

Gemini Keys (Commas/Newlines for Shuffle & Failover)

⚡ Pre-Upload & Pre-Stage Pipeline (~3.5s Generation): Selecting a file immediately downsamples to 16kHz mono in 30ms and pre-stages Deepgram Nova-3 STT in the background. Clicking "Transcribe" generates Gemini captions in ~3.5 seconds!

Pass 2 Gemini Refiner

Output Script / Format

Spoken Language

Max Words per Line

Highlight Names & Sudden Expressions (e.g., "Zara", "Aiyo!", "Wow!") Smart Emojis (Relevant Emotions & Key Nouns Only)

Select Audio or Video File (MP4, MP3, WAV, M4A, WebM, MOV)

0%

Live Zero-Lag Media Caption Stage (60 FPS Engine)

Process media to test zero-lag synchronized captions.

100% Precision Word Timestamps (Click row to seek media)

#	Word	Start (ms)	End (ms)	Duration (ms)	Type / Flags	Formatted Time
No refined data generated yet.

Full Refined Transcript

Deepgram Nova-3 Raw JSON Output (Pass 1 Baseline)

// Deepgram JSON will appear here...

Gemini Raw JSON Output

// Gemini JSON will appear here...