The Future of Video Is Text: I Built My Own Adobe Premiere Extension
Using vibecoding tools, I built La Subtituleta, an Adobe Premiere Pro extension designed to solve a surprisingly stubborn problem: generating accurate Spanish captions.
How it was made
I was tired of spending up to half an hour captioning every video, only to fix many of the mistakes in Premiere Pro’s automatic Spanish transcripts. So I used ChatGPT and a vibe-coding workflow to build an extension, tested and refined it on real projects for two months, and then released it for anyone to use—for free.
The Result
La Subtituleta runs inside Premiere Pro and uses AssemblyAI, Speechmatics, or Deepgram to generate more accurate captions. It creates both an SRT file, ready to use in your timeline, and an enriched JSON file containing timestamps, speaker labels, and additional transcript metadata. It can also reformat captions for horizontal or vertical videos without sending the audio to the API again.
The JSON turns every video into structured text that Claude, ChatGPT, Codex, or another LLM can analyze. This makes it easier to find key moments, create first drafts of social media clips, and build a searchable archive—using far fewer tokens than uploading the entire video file.
The goal is simple: spend less time fixing captions and more time editing and making better videos.
You can see a short video about the project HERE. A full tutorial HERE. Download the tool HERE.
NICORUSSO