Dictate in the language you're already typing in — VTT detects your keyboard layout and routes to the right engine across 26 languages. Speak; text appears at your cursor in any Mac app. On-device or cloud engines, your call. No account.
Finally the apple verification is done, and I signed the app) now it can be used by other people without any pain =D Wispr, it is time to start to be scared
No reviews yetBe the first to leave a review for VTT for Mac
Maker
📌
Hey PH! I made VTT because I live in two languages, and dictation apps quietly assume you don't.
Especially when they starting to translate FOR YOU! But I HAVE NOT ASKED!
Most tools give you one engine for everything. On-device recognition is great for English but weaker for smaller languages — so you either accept bad transcripts, or ship all your audio to a cloud server, including the parts that never needed to leave your Mac.
VTT lets you decide per language: keep English on-device, route your second language to Deepgram, OpenAI, or ElevenLabs with your own API key. 26 languages, and it follows your keyboard language automatically.
It's free, no account, built solo — so honest feedback shapes the roadmap directly. Which language pairs do you dictate in, and which engine handles your non-English language best?
Report
The auto-routing per keyboard layout is such a thoughtful detail. Every dictation app I've tried either forces one engine for everything or makes me switch manually, and it always breaks mid-thought. Keeping English on-device while routing a second language to a cloud engine automatically is exactly how I'd want to configure this, and I'm genuinely glad someone built it.
Report
Maker
@matthieu_poitrimolt That's exactly the bet — manual switching kills the flow, so the routing has to be invisible. On-device for the dominant language, cloud only where it earns its latency. Glad it clicked for you.
Report
Love that English stays local. One thing though: the second language is usually the more personal one, and that's the bit going to the cloud. Do I get a clear signal in the moment that audio is leaving my Mac, or is it set once and then invisible?
Report
Maker
@angelika_dev You can keep it fully on-machine. Model choice is per language — set your second language to a local model and it never leaves your Mac, exactly like English. The cloud is only used for a language if you pick a cloud model for it. So nothing personal has to go out at all; it's a per-language choice, not one global setting you're stuck with.
How's it different than so many other apps in the category?
Report
Maker
@syed_shayanur_rahman Quick question: why should the model guess what language you're speaking when your keyboard already knows? That's the pain behind VTT...
The auto-routing per keyboard layout is such a thoughtful detail. Every dictation app I've tried either forces one engine for everything or makes me switch manually, and it always breaks mid-thought. Keeping English on-device while routing a second language to a cloud engine automatically is exactly how I'd want to configure this, and I'm genuinely glad someone built it.
@matthieu_poitrimolt That's exactly the bet — manual switching kills the flow, so the routing has to be invisible. On-device for the dominant language, cloud only where it earns its latency. Glad it clicked for you.
Love that English stays local. One thing though: the second language is usually the more personal one, and that's the bit going to the cloud. Do I get a clear signal in the moment that audio is leaving my Mac, or is it set once and then invisible?
@angelika_dev You can keep it fully on-machine. Model choice is per language — set your second language to a local model and it never leaves your Mac, exactly like English. The cloud is only used for a language if you pick a cloud model for it. So nothing personal has to go out at all; it's a per-language choice, not one global setting you're stuck with.
TestMu AI
How's it different than so many other apps in the category?
@syed_shayanur_rahman Quick question: why should the model guess what language you're speaking when your keyboard already knows? That's the pain behind VTT...