Five export formats. One for every workflow.

Vocateca exports every transcript as Markdown, plain text (.txt), SRT, HTML, or OKF (Open Knowledge Format). Use Markdown for your Obsidian or Notion archive, SRT for subtitles, HTML to share one readable file, plain text when a tool just needs the words, and OKF when a script or LLM has to read the transcript.

Compare the formats

All five are generated straight from the same transcript, so timestamps and speaker labels — where a format supports them — always match the source audio.

FormatBest forTimestampsSpeakersMetadata
.md
Markdown
Archive · Obsidian & NotionNo — pair it with the .srt sidecar for cuesYes, as bold headers in the bodyFull YAML frontmatter (guid, show, title, publish date, source URL, detected language, engine, word count, duration, provenance) plus an Obsidian wikilink to the show note
.txt
Plain text
Quick paste · diffing · LLM promptsNoNoNone — frontmatter and headings are stripped
.srt
SRT
Subtitles · video editors & playersYes, one per caption cueYes, [S1]/[S2]-style prefixes when detectedNone
.html
HTML
Sharing as one self-contained fileNoNoTitle, show, and publish date in the page header only
.okf.md
OKF
LLM ingestion · programmatic use · knowledge toolsYes, MM:SS per segmentYes, "Speaker N" labelsMinimal YAML frontmatter (type, title, resource URL, source, speakers, tags, timestamp, provenance)
Open format

What is OKF?

OKF (Open Knowledge Format) is an open, minimally-opinionated Markdown + YAML convention for portable knowledge documents, based on the public knowledge-catalog specification — not a Vocateca invention or a lock-in format. Turn on "Save .okf" in Settings → Export and every transcript also gets a matching .okf.md sidecar next to the regular Markdown file.

The frontmatter stays small on purpose — type, title, the source URL, detected speakers, tags, and a timestamp — while the body timestamps every line down to the second. A script or LLM can index it without learning Vocateca's own Markdown conventions first.

Example .okf.md

---
type: reference
title: "How On-Device Transcription Works"
resource: "https://feeds.example.com/ep-042.mp3"
source: "podcast"
speakers: ["Speaker 1", "Speaker 2"]
tags: [transcript, my-show]
timestamp: "2026-05-14T09:00:00Z"
origin: "asr:whisper:large-v3"
---

# How On-Device Transcription Works

## Transcript

**[00:00]** Speaker 1: Welcome back to the show.
**[00:04]** Speaker 2: Thanks for having me — good to be here.

Where exports fit

LLM workflows

Feed OKF or plain-text transcripts straight into a RAG pipeline, a prompt, or an agent tool call — no HTML tags or app-specific Markdown quirks to strip out first. The vocateca-cli library export --format okf command pulls one straight to a file or stdout for scripting.

Read about LLM workflows

Obsidian & Notion

Markdown exports carry a YAML frontmatter block — guid, show, publish date, detected language, word count, duration, and provenance — plus a wikilink to the show note, so Obsidian treats every transcript as a linked, searchable page the moment it lands in your vault. Notion works differently: Vocateca can push transcripts straight into a Notion database through its own integration, separate from this file export.

Try it on your own transcripts

Download for macOS