Audio in.
More possibilities out.

One recording creates a lot of follow-up work.

Sharing audio as video involves more than an export. The recording needs a transcript, readable captions, the right aspect ratio, and supporting assets. Repeating that process by hand makes each new recording another production task.

A connected workflow from recording to output.

I built an automated media pipeline with a React interface and Flask backend. OpenAI Whisper provides transcription, and the Python pipeline produces portrait and landscape videos with synchronized subtitles and supporting assets.

The interface brings the main steps together: upload the recording, transcribe and review the text, configure the output, and render the result.

1. 01

   Upload

   Start with an audio recording.

2. 02

   Transcribe & review

   Turn speech into timed text, then review it.

3. 03

   Configure

   Set up the caption and output presentation.

4. 04

   Render & deliver

   Produce video, subtitles, thumbnails, and metadata.

Connect media processing to a usable interface.

The central engineering challenge is coordination: text, timing, layout, and output format all need to agree. A correction to a transcript should be understandable in the interface and remain aligned with the media.

What synchronized captions make possible.

The interactive illustration lets you explore word highlighting and portrait versus landscape layouts. It uses a fixed sample transcript and illustrative timing, so you can try the presentation immediately in your browser.

From one recording to a set of deliverables.

The application produces captioned videos and supporting assets. This portfolio documents its workflow; a public hosted version and source repository are not linked here.

For a conversation about the implementation, get in touch.