Translate your OBS Studio stream live

OBS Studio is the broadcast brain of most independent streams. It mixes scenes, encodes video, and pushes a single stream to Twitch, YouTube Live, Kick, or a custom RTMP target. What it does not do well is translate. OBS’s audio graph is built for the broadcaster’s perspective: a single mix going out to a single audience.

A streamer with an international audience faces a familiar tension. Adding interpretation through a third-party plugin compromises the encoder. Re-streaming to language-specific channels splits the chat and dilutes engagement. Most creators end up doing nothing — and watching non-English viewers drift away after a few minutes of half-following the stream.

Loquira runs in parallel. It never sits in the OBS audio chain, never re-encodes your stream, never asks for a virtual cable on your operating system. The result is that adding translation to an OBS stream is a five-minute change — and if you turn it off, your stream is exactly what it was before.

How the parallel model works

The microphone you already use for OBS produces an analog signal. That signal can be read by more than one application at the same time. OBS reads it for the stream. Loquira reads the same input from a second device — a phone, a tablet, or a second laptop — running the Loquira presenter view.

You speak once. OBS captures your voice for the live audience that already speaks your language. Loquira captures the same voice and transcribes it, translates it into the languages your international viewers selected, and synthesises speech in those languages. Each non-English viewer holds their phone, scans the QR code on your stream overlay, picks their language, and listens through earbuds while watching the video on the main screen.

Recommended setup

• Microphone. A wired condenser or dynamic mic on a boom — see choosing the right microphone for what works best for solo streamers. USB mics work; XLR-through-an-interface is better for sustained sessions.
• Mic position. Close (15 cm or less). Loquira’s recognition accuracy degrades quickly in reverberant rooms — our audio requirements page lists the thresholds.
• Second device for Loquira. A phone running the Loquira presenter view is the simplest setup. Mount it on a small tripod beside your mic; the phone’s built-in mic does not need to be the input — pair it with the same audio interface OBS uses, or position it close enough to capture the same source.
• QR code on stream. Export the Loquira join QR as a PNG and add it as an image source in one of your scenes — ideally a “starting soon” or “be right back” scene where chat naturally pauses to look at the screen. A small persistent corner overlay also works.

What about translating chat or alerts?

Loquira translates speech, not text. Stream chat translation belongs to your chat overlay or a moderation bot — keep them separate. Alerts (donations, subs) are typically too short to translate meaningfully; let the speaker react to them in their own language and the natural reaction will be translated downstream.

Latency on a stream that is already buffered

OBS streams typically carry a 5–15 second buffer between you speaking and the viewer hearing through the platform’s CDN. Loquira’s translated audio reaches the viewer in roughly 1.5 seconds end-to-end. The result is that translated listeners hear the translation before the main stream audio reaches the buffered viewer. The recommended viewer experience is to keep the main stream muted and listen entirely through Loquira on a phone — the synchronised transcript on Loquira’s listener view replaces the captions overlay.

If you also want viewers to hear your original voice synchronised with the translation, ask them to use a low-latency player (Twitch’s experimental “low latency mode” or YouTube Live’s “ultra low latency”) and to listen to the stream’s audio alongside the Loquira feed at a small volume. This is the same trade-off interpreters at conferences manage — most viewers prefer one clean voice over two layered ones.

Known limitations

• Music and singing. The translation pipeline is tuned for speech. Songs played through the stream will be transcribed as garbled text and translated incoherently. End or pause the Loquira session during dedicated music segments — the QR overlay can stay; the listener will simply hear silence until you resume.
• Two streamers on one OBS instance. If you collaborate with a co-streamer over Discord and pipe their voice into OBS, Loquira will only transcribe the audio it captures locally. Each speaker should run their own Loquira presenter session and share their own QR code — viewers pick whose voice to follow.
• Recordings. OBS recording is independent of the live Loquira session. If you want a translated version of a VOD, run Loquira live during the broadcast — the transcript and language replays are saved and remain accessible after the stream ends.