To translate a Discord voice call into text you can read, use a desktop translator that explicitly captures computer audio. Join the call in Discord, select the spoken and translation languages in that tool, choose computer sound as its input, and start live captions.
The audio source is the deciding detail. A microphone-only translator may hear you clearly while receiving nothing from the people speaking through your headphones. A text translation bot usually solves a different job: written messages in a channel.
Decide what you need the translator to do
| Your goal | Capability to look for | What it does not establish |
|---|---|---|
| Understand other people in voice chat | Computer-audio capture and translated live captions | That other callers hear your reply in their language |
| Have translated speech read aloud to you | Voice output in addition to translation | That this output is routed into Discord’s microphone input |
| Send your speech back in another language | A documented outgoing audio route, or a voice bot with the required channel permissions | That selecting “two-way” automatically configures Discord audio routing |
| Translate text in a server | A text translator app or message command | Live voice-channel translation |
I would set up incoming captions first. You can check whether you understand one speaker before adding spoken output, another language or a more complicated audio route.
A desktop setup for incoming Discord voice
Akkadu’s Discord setup guide provides a concrete example for Windows and macOS. It documents its desktop app working alongside Discord’s desktop app or web version, using computer sound as the source. Check the tool’s current OS requirements, language pair and usage allowance before installing.
- Open Discord and check the ordinary call audio. Join the voice channel or private call. Confirm you can hear another participant through the headphones or speakers you intend to use.
- Install and open the desktop caption tool. For the Akkadu example, sign in, open Languages, and choose Original language for what the other person speaks and Translated language for what you want to read.
- Select computer audio. In Akkadu’s documented flow, open Audio Settings → Sound source and select Computer sound. Approve the OS audio-capture permission if requested. A microphone permission by itself is not the same capability.
- Show the captions you need. Its subtitle settings allow original, translated or both languages. Showing both helps you spot a player name or place name that speech recognition misheard.
- Start the translation and ask for a test sentence. Check that another person’s voice produces captions. Pin or position the caption window where it remains visible beside Discord.
- Stop when the conversation ends. Review the service’s transcript settings. Akkadu’s guide says original and translated transcripts are saved, so know where those records go before using it in a private conversation.
Close unrelated media before starting: a tool listening to computer sound can also encounter music, videos or notification audio. This workflow displays captions on your screen; the service may still process or store conversation data remotely. It does not change the language of your Discord microphone.
JotMe’s Discord guide describes another desktop computer-audio workflow, with spoken-language and translation-language controls. It also describes voice output. Treat those as product-specific features: check the current mode and audio-routing instructions rather than transferring Akkadu’s setting names to another app.
Why captions work but your reply is still untranslated
The incoming path is Discord audio → caption tool → your screen. The normal outgoing path is your microphone → Discord → other callers. Unless you deliberately connect a translation tool’s generated speech to the outgoing path, the other people still hear your original voice.
For a first conversation, use translated captions to understand incoming speech and type short translated replies in the text channel. If spoken replies are essential, choose a tool with explicit instructions for sending its translated audio into Discord on your OS. Test with a willing partner and ask which language they actually hear. Hearing a translated voice in your own headphones is not that test.
Use headphones when generating spoken translations so the sound does not return through your microphone and produce repeated translations or echo.
If Discord audio works but the captions do not
- Only your own speech appears: the translator may be listening to the microphone. Check its computer-audio source.
- You cannot hear the other person either: fix Discord audio first. In User Settings → Voice & Video, check Output Device, volume and whether you are deafened. Discord’s troubleshooting guide also covers individual user volume and channel permissions.
- Audio is clear but no text appears: check the translator’s capture permission, selected audio source, session status and supported language. Restart its translation session after correcting the source.
- The captions contain the wrong words: confirm the spoken language and ask one person to speak at a time. A translation based on a mistaken transcript will carry that mistake forward.
- Game sounds appear in the transcript: reduce unnecessary background audio and keep instructions short. If the tool supports a narrower application source, use its documented selection method.
Make it usable in an actual voice channel
Imagine a gaming group explaining a new route through a map. Agree on the map’s location names in text before the round. During the explanation, ask for one instruction at a time and use a short check: “Wait at the north gate, then move when you say go?” The translation has a concrete decision to support.
Live captions can arrive after the sound and change while a sentence is still being recognized. Do not make a fast timing call depend entirely on reading a translated paragraph. For a team discussion or community meeting, leave pauses and put final decisions in the text channel.
Tell participants when a third-party service is processing or retaining the conversation, and choose transcript settings that fit the group. Desktop software can capture audio locally while still using a remote service to recognize or translate it; capture location alone does not prove local-only processing.
Owll Translator’s browser voice tool currently takes live microphone input and displays translated text captions. It does not capture Discord, a browser tab or system audio, and it does not produce spoken output. That makes it a different tool from the computer-audio workflow above. Keep the input and output requirements visible when comparing any Discord voice translator.
