The "dictate, convert, refine" workflow
Record the long-form note as audio (phone's voice memo app, dedicated dictation hardware, anything that produces an MP3 or M4A). Open Audio to Markdown, upload, click Convert, download the .md. Open in VS Code, hit Ctrl+Shift+V for the side-by-side preview. Edit normally — fix transcription errors, add headings, restructure sections, drop in code blocks. The starting point is your spoken thought; the final artefact is polished prose.
For multi-section content, the converter's structure preservation matters: if you said "first, the architecture; second, the data layer; third, the API", the converter often picks up the implicit sectioning and inserts headings. You then refine in VS Code rather than starting from a wall of text.
Why this beats VS Code's built-in dictation
VS Code has limited dictation extensions (mostly third-party, often laggy, mostly single-language). Recording-then-converting is two steps, but each step is robust: any audio capture you trust + a structured Markdown output. Multi-language support, speaker labelling for paired-programming sessions, and offline-friendly source files all come for free.
Pair with PDFs (PDF to Markdown for VS Code) and URLs (URL to Markdown for VS Code) — the same Markdown preview workflow handles all three input types.
Long-form thought capture
The killer use case: writing things you know you should write but never get around to. Architecture overviews, runbooks, postmortems, design proposals. Speak the first draft in 20 minutes, convert, refine in VS Code in another 30. Total time: under an hour for a document that would have taken half a day starting from a blank file.