Adding live captions to OBS comes down to three options: a free local plugin that runs on your GPU, a cloud service like StreamTranslate at $14.99/month, or a browser tab that requires no install at all. MirrorCaption is the third option — a browser-based real-time transcription and translation tool covering 50+ selectable languages, with pricing that starts free and tops out at €99 one-time for Premium.
You're 40 minutes into a live Twitch set. You switch from English into Japanese for your regulars. Half your chat is in Katakana. The other half types "what did you just say?" Your OBS scene has no captions. There's no translation layer, no way for the English-speaking side of the audience to follow along. The moment passes.
MirrorCaption captures your microphone in a Chrome browser tab running alongside OBS. As you speak — in English, Japanese, Korean, or wherever the conversation goes — the transcript and translation appear in real time, side by side. This page covers how to integrate it with your streaming setup and where it fits compared to the alternatives.
Key Takeaways
- OBS has no built-in captioning engine; all live caption solutions require an external tool.
- MirrorCaption runs in a Chrome browser tab alongside OBS — no plugin to install, no GPU drain on your streaming machine.
- 50+ selectable languages with GPT-based translation, covering Japanese, Korean, Arabic, and Hindi — well beyond StreamTranslate's 28 languages.
- Three practical workflows: personal monitoring on a second display, OBS browser source overlay, and real-time translation of source content via Meet mode.
- Premium is €99 one-time with 200h of hosted transcription credit included; at StreamTranslate's current pricing, the equivalent annual cost is roughly $180.
Why Live Captions Matter for OBS Streamers
OBS Studio ships without any transcription or captioning engine — it's a capture and encoding tool, not a speech recognition platform. Every caption layer you see on streams comes from a third-party tool added on top.
The gap matters more than it used to. According to TwitchTracker's language distribution data, the United States accounts for roughly 20% of Twitch viewership. The other 80% of your potential audience is distributed across Germany, Korea, Russia, France, and dozens of other countries — many of whom will follow your stream more easily with captions in their own language.
Three things live captions do for a stream:
- Accessibility. Deaf and hard-of-hearing viewers can follow along. Many viewers also watch with audio off in shared spaces.
- International reach. Translated captions let viewers from other countries follow content in the language they read fluently.
- VOD discoverability. Captions burned into video — or attached as a searchable transcript — give Twitch and YouTube more content to index for search.
The challenge is that most caption tools make you choose between a monthly subscription or GPU-heavy local software on the same machine doing video encoding. MirrorCaption offers a different path.
Bilingual Streamer (JP/EN)
Switches between Japanese and English mid-stream. MirrorCaption shows both languages side by side on a second monitor, so the streamer can verify translation accuracy in real time before viewers notice a mistranslation in chat.
International Audience Growth
An English-speaking streamer adds translated captions for Korean, Spanish, and German viewers. Captions appear as an OBS browser source overlay — same stream, multiple languages readable by different audience segments.
Reaction Streamer
Watching foreign-language content on stream. MirrorCaption's Meet mode captures the source tab's audio and translates it in real time, letting the streamer react authentically rather than guessing from visual context.
Accessibility-First Channel
Captions for deaf and hard-of-hearing viewers displayed directly on stream. No Twitch-side settings required from the viewer — the caption overlay is burned into the video.
Three Ways MirrorCaption Works With OBS
MirrorCaption is a Progressive Web App. It runs in a Chrome browser tab, captures audio via your browser's microphone or tab-capture APIs, and streams transcription and translation back to the screen in real time. It runs alongside OBS, not inside it. There are three practical setups.
Approach 1 — Personal Monitoring on a Second Display
The simplest setup, and the most common starting point for bilingual streamers.
- Open
mirrorcaption.com/appin Chrome on your streaming machine. - Select Talk mode. This captures your microphone input.
- Choose your source language (what you're speaking) and your target language (what you want translated).
- Start the session and move the MirrorCaption tab to a second monitor.
- Your speech appears as a live transcript with side-by-side translation as you stream.
A Korean-English streamer uses MirrorCaption on a second display to monitor how Korean honorifics are being translated before an excited moment gets clipped and shared out of context. The side-by-side view lets them rephrase on the fly when the translation looks off.
Viewers don't see MirrorCaption in this workflow. It's a personal monitoring layer — like a real-time confidence monitor showing you exactly what your bilingual audience is reading.
Approach 2 — OBS Browser Source Overlay
To show live captions visibly on stream, add MirrorCaption as an OBS browser source.
- Open MirrorCaption in Chrome and start a Talk mode session.
- In OBS, add a new Source → Browser Source.
- Enter
https://www.mirrorcaption.com/appas the URL. - Set the width and height. The full app UI renders as a layer.
- Use the OBS Crop/Pad filter or Transform settings to isolate the live transcript area, then position it at the bottom of your scene.
A Japanese streamer adds MirrorCaption as a browser source, crops it to show only the translated English caption line, and pins it to the lower third of their gameplay scene. International viewers can follow the English translation while Japanese-speaking viewers read the original in chat.
One honest note: tools like StreamTranslate generate a dedicated caption-strip URL that shows only the caption text with no surrounding UI — which is slightly cleaner to set up as an OBS overlay. MirrorCaption's browser source requires a one-time crop step to achieve the same result, but OBS saves that configuration in your scene collection, so setup only happens once.
Approach 3 — Real-Time Translation of Source Content (Meet Mode)
This workflow is purpose-built for reaction streamers watching foreign-language video.
- Open the content you're reacting to (a YouTube video, a foreign stream, a video call) in a Chrome browser tab.
- Open MirrorCaption in a second Chrome tab and select Meet mode. This captures the audio from the content tab rather than the microphone.
- Set the source language (the language of the content) and your translation target.
- Start the session. MirrorCaption translates the content audio in real time on your display.
An English-speaking streamer reacts to a Japanese variety show. Running MirrorCaption in Meet mode on a second display, they see translated captions appear as the show plays, letting them respond to what's actually being said — not just what they can read from facial expressions and sound effects.
OBS Caption Tools Compared
Four tools currently recommended most in the OBS streaming community:
| Dimension | MirrorCaption | StreamTranslate | LocalVocal | Caption.Ninja |
|---|---|---|---|---|
| Price | Free trial (1h); €99 one-time Premium | From $14.99/month | Free (open source) | Free |
| Languages | 50+ | 28+ | ~100 (Whisper-based) | Browser-dependent |
| OBS plugin needed | No | No | Yes | No |
| GPU usage | None (cloud) | None (cloud) | High (Whisper local) | None |
| Translation quality | GPT-based | Multi-engine AI | M2M100 model | Browser ML |
| Dedicated overlay URL | Full app UI via browser source | Caption strip URL (cleanest) | OBS text source | Caption strip URL |
| React-stream translation | Yes (Meet mode) | No | Complex manual setup | No |
StreamTranslate is the most purpose-built OBS overlay tool available. Paste one URL into OBS, and a clean caption strip appears. If the major European and Asian languages in their 28+ set cover your audience, and you're comfortable with a monthly subscription, it's the path of least resistance.
LocalVocal is the right pick if you want fully offline, privacy-first processing and have a capable GPU. Be aware it requires manual installation, can break on OBS major version updates, and adds meaningful GPU load to a machine already encoding video.
MirrorCaption wins on language depth, translation quality for Japanese, Korean, and Arabic, total cost over time, and the reaction-stream translation use case — at the cost of a slightly more involved initial browser source setup compared to StreamTranslate's dedicated strip URL.
Stream in Any Language
Start with 1 free hour. No credit card. No OBS plugin to configure.
Open MirrorCaption Free →50+ Languages — Including the Ones Your Audience Actually Speaks
StreamTranslate covers 28+ languages. LocalVocal runs Whisper locally and covers more languages in raw count, but translation quality varies considerably. For Japanese, Korean, Arabic, and Hindi, a local neural machine translation model produces notably different output than GPT-based translation with contextual awareness of the surrounding conversation.
That gap is most visible during live streaming. Bilingual streamers who code-switch between Japanese and English — a common pattern in Japanese gaming streams — need a translation engine that handles mid-sentence language switches, casual register shifts, and gaming slang simultaneously. An approach that predicts words in isolation struggles with that. GPT-based translation, fed the previous several lines of conversation as context, handles it considerably better.
MirrorCaption also shows the original transcript and translation simultaneously, with word-level linking: tap any translated word to see which source word it maps to. For bilingual streamers who want to spot-check their translation accuracy during a live broadcast, that feature is a real-time audit tool.
For streamers building an international audience beyond European languages, the 50+ language set covers Mandarin, Japanese, Korean, Arabic, Hebrew, Hindi, Russian, Vietnamese, Indonesian, Thai, and more — through the same interface and the same one-time pricing.
No Plugin, No GPU Drain, No OBS Update Anxiety
One of the most consistent complaints about LocalVocal in OBS forum threads is version compatibility — OBS major releases have broken the plugin's audio filter chain, requiring manual reinstallation or waiting for a plugin update. For streamers who rely on captions as part of their accessibility setup, a broken plugin on stream night is a genuine problem.
MirrorCaption runs entirely outside OBS. It lives in a Chrome browser tab. An OBS update can't break it. A GPU driver update doesn't affect it. It has no interaction with OBS's audio filter chain, plugin registry, or encoder settings. The configuration that works today will still work after the next three OBS releases.
There's also a performance angle. Local AI transcription adds meaningful GPU load to a machine already encoding video at 1080p60 or above. Streamers on mid-range hardware sometimes report dropped frames when a local transcription model competes with OBS for GPU memory. MirrorCaption offloads all speech processing to the cloud, leaving the GPU free for encoding.
The Subscription Math
At StreamTranslate's current pricing of $14.99/month, a year of streaming with captions costs roughly $180. Two years costs $360. The cost compounds with every month you stream.
MirrorCaption Premium is €99 one-time. It includes 200 hours of hosted transcription credit upfront, permanent product access, and all future updates. Voice Packs (sold separately) top up hours when the included credit runs out; Premium customers get the lowest per-hour Voice Pack rate. At $14.99/month in equivalent terms, MirrorCaption's cost is covered after roughly seven months of streaming — after which every subsequent stream is at no additional cost.
For streamers who don't broadcast every day — or who want caption support for specific events rather than every stream — the one-time model removes the "am I getting value this month?" calculation entirely. Start with the free 1-hour trial, no credit card, to check the translation quality on your content before committing.
Frequently Asked Questions
Does OBS have built-in live captions?
No. OBS Studio does not include a built-in captioning or transcription engine. Streamers add live captions through external tools — either browser-based services, OBS plugins, or dedicated overlay generators. MirrorCaption is a browser-based option that requires no OBS plugin.
Can I add MirrorCaption as an OBS browser source?
Yes. Open the MirrorCaption app in Chrome, start a Talk mode session, then add the tab URL as a Browser Source in OBS. Use the Crop/Pad filter or Transform settings to isolate the live caption area and position it over your scene. OBS saves this configuration per scene, so the setup only happens once. For a cleaner dedicated overlay URL approach, StreamTranslate generates a caption-strip URL that requires no cropping.
Do live captions appear in Twitch VODs?
Browser source overlays render as visual elements burned into the stream video — they appear in VODs and clips as part of the video frame. This is different from CEA-608 embedded captions, which toggle via the Twitch player's CC button and require a separate encoder plugin like the OBS Closed Captioning plugin. MirrorCaption's overlay approach means captions are permanently visible in the recorded video without viewer-side settings.
Which languages does MirrorCaption support for streaming?
MirrorCaption supports 50+ selectable languages, including Japanese, Korean, Mandarin, Arabic, Hindi, German, Spanish, French, Portuguese, Russian, Hebrew, Vietnamese, Indonesian, Thai, and more. Translation uses GPT-based refinement, which handles code-switching and conversational speech better than dictionary-based or isolated-word approaches. See the multilingual transcription guide for a language-by-language breakdown.
What is the difference between MirrorCaption and LocalVocal?
LocalVocal is a free OBS plugin that runs AI transcription locally on your GPU — a good fit for privacy-first, offline setups with capable hardware. MirrorCaption is browser-based: no OBS plugin, no GPU usage, no version compatibility risk. MirrorCaption's GPT-based translation handles Japanese, Korean, and Arabic with higher contextual accuracy for most live streaming use cases.
Is there a free trial for MirrorCaption?
Yes. Every account starts with 1 free hour of live transcription and translation — one-time, no monthly reset, no credit card required. Upgrade to Annual (€54.99/year, 100 hours of hosted transcription credit included) or Premium (€99 one-time, 200 hours included, all future updates and new features with priority access) when you need more hours. Voice Packs (sold separately) add credit on any plan at any time; Premium customers get the lowest per-hour rate.
Add Live Captions to Your Next Stream
1 free hour to try. No credit card. No plugin to install. Works in Chrome alongside OBS.
Try MirrorCaption Free