Can you transcribe a Discord call or voice chat?

Yes. You can transcribe a Discord call or voice channel in one of three ways: add a bot that joins the channel, capture the audio on your own computer with a desktop app, or record the call and transcribe the file afterward. We could not find a Discord help article that describes a transcription feature for voice channels, so check Discord's help center before you assume the app has one. The method you choose decides whether other people see an extra participant, where the audio goes and how much setup you need.

A bot can listen to a voice channel because Discord's voice protocol works both ways

A bot is the only method where the software takes part in the call. Discord's developer documentation says that voice data is carried over UDP in both directions, and that a client must be able to receive UDP packets ¹. That two-way connection is what lets a bot hear the people in a channel. A bot shows in the member list, so everyone can see that something is listening.

Bot directories list many tools that advertise call transcription. Before you add one, check three things. First, who runs the service. Second, where it sends the audio: a bot you host yourself keeps the audio on your own server, while a hosted bot sends it to the company that runs it. Third, what its privacy policy says about how long it keeps recordings. A listing is not evidence of accuracy, so test any bot on a short call before you rely on it.

Capturing your computer's audio needs no extra participant

A desktop app can record the audio your computer plays, including the Discord call, without joining the channel. Meetily, an open-source meeting app, says it captures microphone and system audio at the same time ². Its README does not say which systems support system audio capture, so test it before you rely on it. If you want the app to stay on your own machine, read the summary settings first, because Meetily lets you send summaries to cloud providers ³. Our open-source meeting transcription question compares the tools in more detail.

On a Mac, Apple's Live Captions can caption the computer's audio or the microphone, and you can switch between the two ⁴. The feature runs only on Mac computers with Apple silicon ⁵. Live Captions shows text as people speak. The Apple pages we checked describe captions on screen, not a saved transcript, so test whether you can keep one before a call that matters.

A recording you transcribe afterward gives you the most control

If you need a file, record the call with a tool you trust, then transcribe the file. You can check the finished transcript against the audio at each point that matters, and you can choose the model. The steps are in our guide to how to transcribe audio to text. For a short file, the free audio to text tool runs in your browser. If your recording is a voice message rather than a call, see how to transcribe voice messages.

Consent comes before the first word is captured

Tell everyone before the recording starts. The law can require it. In the United States, the federal Wiretap Act allows a recording when one party to the conversation consents, which is the one-party rule ⁶. California Penal Code 632 is different: it punishes recording a confidential communication without the consent of all parties ⁷. If anyone on the call is in California, get everyone's consent. The terms are defined in the all-party consent and one-party consent entries.

A bot makes the recording obvious to everyone in the channel, and an announcement in the channel covers the rest. Say that the call is being transcribed, where the transcript will be stored and who can read it. If someone objects, stop the recording. Our guide to meeting transcription without a bot covers the same consent steps for other platforms.

What to do

1. Decide whether you need live text on your screen or a saved record. Live captions are not a record unless a tool saves them.
2. Choose a method from the table. Use a bot if everyone should see it and you trust its host. Use a desktop app if you want no extra participant.
3. Announce the transcription in the channel before the call starts, and ask anyone who objects to say so.
4. Run a short test call, then listen for dropouts and muffled voices. Overlapping speakers lower accuracy in every method, as the meeting transcription guide explains.
5. After the call, check the transcript against the audio at each decision, and keep the audio if the record matters.

For the wider picture, see the meeting transcription guide and our transcription overview.