Using AI for daily tasks: Transcript Audio Example

Transcript Audio Script

Lately, we hear “AI agent” and “workflow” everywhere. You’ve probably seen tools like n8n or Zapier. They’re powerful, but for beginners, their interfaces aren’t exactly intuitive. Especially those canvas-style UId. I think they have an unnecessarily high learning curve.

Sometimes, when I need something, I just build my own small tools. Instead of dealing with heavily marketed products, I prefer making simple, goal-oriented solutions.

The repo I’m introducing today is one of those: Transcript Audio.
GitHub Repo: sedyldz/transcript-audio

Why I Built This

There are thousands of services for transcription. But now, with local models, you can get high-quality results without sending your data anywhere, and do it for free.

I’m sharing this repo for those who want to “do it themselves.” I believe the future is moving toward:

• Small AI agents that solve very specific needs
• Just enough coding knowledge to modify them
• Enough technical skills to manage your own tools

How I Developed It

I didn’t write a single line of code in this repo, but I designed the whole process:

• Converting video to audio, then audio to text in separate steps
• Exporting in different formats
• Accepting multiple videos
• Iterating the process when results weren’t good enough

At first, I didn’t know all the requirements. But as I tested with my own videos, I noticed what was missing and improved it step by step.

First Thing You Should Learn: Python

If you want to start working with AI, I think you should learn these basics first:

1. How to run Python code
2. Where the results are saved on your computer
3. How to set up a virtual environment
4. Using Git to track what changed

Once you have this foundation, you can create highly tailored AI agents for your own needs. The key is being able to iterate on the code.

Conclusion

This repo may look like just a “free audio transcription tool,” but it’s also an example of a bigger idea: building your own AI workflow. As long as your computer isn’t too weak, you can run this model locally without problems.

👉 Check out the repo and take your first step toward creating your own agent.