Learn how to effectively code up an internal tool using Claude Code or Codex. The post Build a Speaker-Recognition App with Claude Code appeared first on Towards Data Science.

One great advantage of having access to coding agents such as Claude Code or Codex is that you can create internal applications super quickly. Before coding agents, you had to spend weeks, if not months, to create internal tooling that was effective for you. You can now do it in a matter of minutes. This has greatly changed the domain of internal tooling.

I saw the need for internal tooling where I hand in a meeting recording audio, and it outputs the person speaking in each sentence. Normally, when you give audio to a meeting transcription tool, it gives you something like what you see below, where it identifies each individual speaker in the recording and labels them as speaker 1, speaker 2, speaker 3, and so on. However, what you really want is to have the name of who says what in the meeting: not only anonymous speaker labels, but actually named speaker labels.

This is the application I built for myself in just a matter of two hours. In this article, I'll discuss why I came up with the idea and how I implemented it. I'll discuss why I needed the speaker recognition app, how we can build internal tooling with Claude Code in general, and how the details of how the speaker recognition app works.