The Format Question

When I was building My Transcriber, one of the first decisions was output format. Where does the transcribed text go?

Most transcription tools store your text in their own database. Some export to Word documents. Some give you SRT subtitle files. Some keep it locked inside their app.

I chose markdown. Plain .md files in a regular folder on your filesystem.

This wasn't an arbitrary decision. It was the most important design choice in the whole app. Here's why.

A File You Own

A markdown file is a text file. That's it. It's not a proprietary format. It's not stored in a database. It's not locked behind an API.

It sits in a folder on your Mac. You can move it, copy it, rename it, edit it, or delete it with Finder. No app needs to be running for you to access your transcriptions.

If My Transcriber disappeared tomorrow, your transcriptions would still be there. Every single one. Readable in any text editor on any operating system.

That's not true of most transcription tools. If Otter.ai shuts down, what happens to your transcriptions? If a note-taking app changes its export policy, do you still have your data?

With markdown files, you don't have to ask those questions. The files are yours.

Opens in Everything

Markdown files open in:

• Obsidian — where they become linked, searchable notes
• VS Code — with syntax highlighting and preview
• Apple Notes — by importing or copying
• TextEdit — it's just text
• iA Writer — beautiful writing environment
• Bear — if that's your thing
• Notepad — on Windows, if you switch platforms
• vim, nano, emacs — for the technically inclined
• Any text editor on any platform — because it's text

You're not locked into one app. If you switch from Obsidian to something else next year, your transcriptions come with you. No export step. No migration tool. Just point the new app at the same folder.

Searchable with Tools You Already Have

Because markdown files are plain text, every search tool on your Mac already works with them.

Spotlight: Type a keyword in Spotlight, and it'll find the voice memo where you said it. Spotlight indexes text files automatically.

Finder search: Open Finder, search in your transcription folder. Instant results.

grep/ripgrep: If you're comfortable with the terminal, you can search with pinpoint precision. Find every transcription where you mentioned "budget" or "deadline" in seconds.

Obsidian search: Full-text search across all your notes, including transcriptions. Link them to other notes. Use tags. Build a personal knowledge base.

You don't need a special app to search your transcriptions. You already have the tools.

Shareable via Any Cloud Sync

A markdown file in a folder can be shared with anyone through any file sync service:

• Dropbox — put the output folder in Dropbox, share with your team
• Google Drive — same idea, works with Google Workspace
• SharePoint / OneDrive — for corporate environments
• iCloud — for sharing across your own devices
• Syncthing — for privacy-conscious peer-to-peer sync

There's no special sharing feature needed. No invite system. No permissions UI. It's a file in a folder. If you can share a folder, you can share transcriptions.

This is why My Transcriber supports team folders. Everyone points to the same shared folder. Each person gets their own subfolder. The transcription files just appear.

Version Controllable

If you use git (or any version control system), markdown files work perfectly.

You can track changes to your transcriptions over time. See what was edited. Diff two versions. Roll back if needed. Collaborate through pull requests.

This might sound niche, but it's powerful. A research team could have all interview transcriptions in a git repo, with annotations and corrections tracked as commits.

Try doing that with a proprietary transcription format.

AI-Ready

This is the one that matters most right now.

Every major AI tool accepts text files as context. Claude Projects lets you upload files. ChatGPT accepts file attachments. Local LLMs read text files directly.

A folder full of markdown transcriptions is immediately usable as AI context. Drop your transcription folder into a Claude Project and you can ask questions about everything you've said this month. "What did I say about the Q2 budget?" "Summarize my voice memos from last week." "Find all the action items I mentioned."

This doesn't work if your transcriptions are locked in an app's database. It doesn't work with proprietary formats. It works with text files.

Your voice memos become fuel for AI tools. Not because of any special integration, but because markdown is the lingua franca of text.

Frontmatter: Structured Metadata

Each My Transcriber transcription file includes YAML frontmatter — structured metadata at the top of the file:

---
captured_at: "2026-03-08T10:15:00+01:00"
duration: "4m 22s"
language: "en"
source: voice_memo
---

This metadata is machine-readable. Tools like Obsidian use it for sorting, filtering, and display. Scripts can parse it. AI tools can use it to understand context.

But it's also human-readable. You can open the file and see exactly when you recorded it, how long it was, and what language it was in. No need to check file properties or a separate database.

And because it's in the file itself, the metadata travels with the file. Copy it to another machine, share it with a colleague, move it to a different folder — the metadata stays.

Future-Proof

Markdown has been around since 2004. It's the de facto standard for plain text formatting on the web. GitHub uses it. Every documentation site uses it. Most knowledge management tools support it.

It's not going away.

Compare that to proprietary formats. How many note-taking apps have come and gone in the last 20 years? How many databases have been orphaned? How many export formats are no longer supported?

Your transcriptions from today will be readable in 2046. Not because any app will still be running, but because text files will still be text files.

That's the kind of guarantee I want for something as personal as voice recordings turned into text. These are my thoughts, my ideas, my meeting notes. They should outlast every app I'll ever use.

The Obsidian Connection

If you use Obsidian, markdown transcriptions are especially powerful.

Each transcription is a note in your vault. You can link it to other notes. Tag it. Add it to a canvas. Reference it in a daily note.

Obsidian's graph view shows how your transcriptions connect to everything else in your knowledge base. A voice memo from a meeting links to the project page. A brainstorming recording links to the feature spec.

This only works because the transcriptions are markdown files in the same vault. Not separate. Not imported. Just files.

What About Rich Formats?

Some people ask: why not Word documents? Why not rich text? Why not HTML?

Because those formats add complexity without adding value for transcriptions. A voice memo transcription is text. It doesn't need fonts, colors, or page layouts. It needs to be readable, searchable, and portable.

Markdown gives you basic formatting when you want it — headings, bold, lists — without the overhead of a binary format. It's the minimum viable format that's still pleasant to read.

If you later need a Word document, converting markdown to docx is a one-line command with pandoc. But you can't easily go the other direction — extracting clean text from a Word document often loses structure.

Start with the simplest format. Convert when you need to. Never the reverse.

What About SRT/Subtitle Formats?

SRT and VTT are great for video subtitles. If you're making YouTube videos or editing podcasts, those formats are essential. Tools like MacWhisper excel at this.

But for voice memos, subtitle formats don't make sense. You're not syncing text to video timecodes. You just want the words.

My Transcriber doesn't output SRT because it doesn't solve the problem it's designed for. If you need subtitles, use a tool built for that. If you need your voice memos as searchable text, markdown is the right format.

One File per Memo

Each voice memo becomes one markdown file. Not a section in a big document. Not a row in a database. One memo, one file.

This means you can:

• Delete a single transcription without affecting others
• Move specific transcriptions to different folders
• Share individual files with people
• Use filenames (which include the date) for quick scanning
• Process individual files with scripts or AI tools

One file per memo is the most flexible unit of organization. You can always combine them later. You can't easily split a monolithic file into individual memos.

The Bottom Line

Markdown isn't exciting. It's not a feature you show off in a demo. Nobody gets impressed by "it writes text files."

But it's the single most important design decision in My Transcriber. Everything else follows from it: portability, searchability, shareability, AI compatibility, future-proofing.

Your voice memos contain your thoughts, ideas, and memories. They deserve a format that belongs to you, not to an app.

Try It

Record a voice memo. Get a markdown file. Own your words forever.

macOS 15+ required. Apple Silicon only. Every transcription is a plain markdown file you own.