Speech, understood.
In any language.
Transcribe and translate live audio with a developer-grade API. Pay only for what you use.
Everything you need to ship voice
A streaming pipeline, honest pricing and the kind of API you actually want to integrate.
Real-time
Streaming WebSocket API. Partial results within 200ms.
200+ languages
Fine-tuned acoustic model paired with neural translation.
Flexible billing
Monthly plans for predictable cost, plus credit packs when you need to scale beyond your quota.
Developer-first
Bearer auth, a clean dashboard and usage metrics.
Capture, transcribe, translate
Three stages between a microphone and translated text — each one streaming.
Capture
Stream audio over a WebSocket from any client — browser, mobile or server.
Transcribe
Varmir Engine returns partial results in ~200ms as you speak.
Translate
The Varmir Engine renders each phrase into your chosen target language.
Pricing at launch
We're finishing the model fine-tuning before opening signups. Join the waitlist and we'll email you when access opens.
- 1,000 min / month — full trial
- All 200+ languages
- Real-time WebSocket + REST
- Email notifications when quota runs low
- 1 active API key
- 8,000 min / month
- Real-time translation
- Usage dashboard
- Email support
- 5 active API keys
- 80,000 min / month
- Priority inference queue
- Unlimited team API keys
- Priority support
- SSO & audit logs
- Unlimited min / month
- Dedicated GPU + SLA
- Self-hosted option
- Dedicated account manager
- Custom model fine-tuning
Pricing is provisional and may change before launch.
Questions, answered
The essentials on how Varmir transcribes, translates and bills.
What is Varmir?
Varmir is a real-time multilingual speech-to-text API with built-in translation. It transcribes live audio and renders each phrase into your target language across 200+ languages, powered by the Varmir Engine.
How does Varmir's speech-to-text API work?
You stream audio over a WebSocket and Varmir returns partial transcripts within about 200 milliseconds as you speak. Each finished phrase is then translated into your chosen target language.
Which languages does Varmir support?
Varmir recognizes speech in 200+ languages and translates between them, including Ukrainian, English, Spanish, German and many more. Language can be auto-detected or set explicitly.
Is Varmir real-time?
Yes. Partial transcription results stream back within roughly 200 milliseconds, and translation follows each phrase with under 200 milliseconds of added latency — suitable for live captioning and dictation.
How is Varmir priced?
Varmir uses usage-based pricing: a free Starter tier, monthly plans for predictable cost, and credit packs to scale beyond your quota. You pay only for the audio you process.
Can I use Varmir for live translation?
Yes. Varmir transcribes speech and translates each phrase into your chosen language in real time, which makes it a good fit for live translation, subtitles and multilingual voice interfaces.
Still curious? Read the docs and API reference, see supported languages and streaming, or check pricing.