How to transcribe your computer's audio in real time

Published September 11, 2026 · 5 min read

The most common question for anyone starting out with transcription is this: "I want to transcribe what's playing on my computer - the other person's voice on the call, the audio from a video - not my own voice on the microphone." It sounds like a detail, but it's the difference between a transcript that's useful and one that isn't.

Microphone vs. system audio

There are two very different sound sources on your PC:

Microphone: picks up your voice and the room around you. It's what most dictation apps use.
System audio (or "loopback"): the sound coming out of the computer - the other person in the meeting, the video's narrator, the player's music. That's the one you want when the topic is a meeting, a class or a video.

Transcribing system audio is what lets you capture the speech of whoever is on the other side, without having to hold a microphone up to the speaker (which would produce error-riddled text).

What it's good for day to day

Meetings: keep a record of what was said on a call, even in another language.
Classes and courses: have the text of a video lecture to review later.
Videos and podcasts: turn spoken content into searchable text.
Accessibility: read what's being said when audio isn't an option.

How to transcribe system audio

On Windows, capturing system audio is possible natively (the operating system exposes that sound stream to apps that know how to ask for it). A transcription program captures that stream, sends it to a speech-recognition AI and returns the text in real time. The typical steps are:

Open whatever you want to transcribe (the meeting, the video, the class).
In the transcription program, choose to capture the computer's audio (not the microphone).
Hit start - the text appears as the person speaks.
At the end, export the transcript to review or archive.

If the content is in another language, you can go beyond transcription and translate it live: the original text and the translation appear side by side, on the spot.

One thing people overlook: privacy

Because it captures your audio, on your computer, no bot joins the meeting as a participant. Nobody on the other side sees a "recorder" on the call. In corporate settings this is often decisive, since recording bots run into internal policies.

Doing it in practice with Lumespeak

Lumespeak captures your computer's system audio and shows the transcript in real time - and, if you want, the translation right next to it. It works with any sound source (Teams, Meet, Zoom, YouTube, a video player) and doesn't join the call. At the end, you export the text.

Want to try with a real meeting or video? Create a free account and give it a go today - no credit card.

Transcribing your computer's audio no longer takes a workaround. By capturing system audio, you turn any sound playing on the PC into text - and, if you need it, into text that's already translated.