Fireflies.ai is great for meeting transcripts, but exporting them as usable Markdown can be messy due to embedded images, tracking pixels, and weird capitalization typos (like doubled uppercase letters at sentence starts).

Here is my workflow for converting a Fireflies transcript into a clean, searchable Markdown file.

1. Get the HTML Transcript

Fireflies doesn’t always provide a direct clean Markdown export. The most reliable way to get all the data is to save the web page itself:

1. Open the transcript: Navigate to the specific meeting in your Fireflies dashboard.
2. Load all content: Scroll through the entire transcript to ensure lazy-loading elements are rendered.
3. Right click to inspect an element, select the html element and then click on “Copy outer html”. Open a text editor and save the content as an html file.

2. Process with Automation

I use a master Python script that chains three specialized tools:

1. Pandoc: Converts HTML to GF-Markdown while stripping raw HTML tags.
2. Image/Emoji Remover: Cleans out data-URIs, tracking pixels, and distracting emojis.
3. Double-Cap Fixer: Fixes common sentence-start typos (e.g., “AAple” → “Apple”).

The Master Script

Run the script on your saved HTML file:

python3 scripts/process_firefly_transcript.py path/to/transcript.html

Source: process_firefly_transcript.py

The script automatically creates a .md file in the same directory, cleaned and ready for your notebook or blog.

3. Manual clean-up

You still need to go through the generated markdown to clean it up. But the effort is significantly reduced.