SUSHI TECH TOKYO 2026 · BOOTH D-1163
Tokyo Big Sight · Apr 27–29

We build your AI Company Bot — live at the booth.

We design and build a custom AI chatbot tailored to your operations. Try the booth build instantly on the web, or come meet us at D-1163 for a 20-minute consultation.

Try the Company Bot
Live company bots
Your bot will appear here once we build it at the booth.
Real-time voice translation

Speak your language.Be understood in theirs.

Maloha translates spoken conversation between 14+ languages in near real time — powered by WebRTC and a stack of live speech models.

Download on iOSGet it on Android
< 500ms
End-to-end latency
14+
Languages supported
6
Live speech models
A
Alex🇺🇸 EN
지우🇰🇷 KO
ENWhere is the nearest subway station?
KO가장 가까운 지하철역이 어디인가요?
ENIt's about five minutes on foot.
KO여기서 걸어서 5분 정도 걸려요.
Core

Built for live conversation, not captions.

Real-time voice

WebRTC transport keeps speech round-trips low enough to hold a natural back-and-forth — no awkward pauses, no push-then-wait.

14+ languages

Korean, English, Japanese, Chinese, Spanish, French, German, Italian, Portuguese, Russian, Arabic, Hindi, Turkish, Dutch — and auto-detect.

Context-aware AI

Multiple models per language pair. The engine picks up idioms, slang, and local register, so the translation sounds like the other person actually talking.

Four translation modes

Conversation, one-way interpretation, learning mode with parallel transcripts, and manual push-to-talk when timing matters.

How it feels

A two-way conversation, not a text box.

Hold to speak in your language. The other person hears it in theirs — and replies the same way. Transcripts stay on-screen so nothing gets lost.

🇺🇸 EN ↔ 🇰🇷 KO
Where is the nearest subway station?
You speak
WebRTC
🇰🇷 KO ↔ 🇺🇸 EN
Where is the nearest subway station?
Who it's for

When the next thing you say needs to land.

Travel

Travel & tourism

Order food, ask directions, sort out a hotel mixup, handle an emergency room — without an awkward silence in the middle of it.

Business

Business & work

International calls, live customer support, multi-language standups, on-the-ground negotiation — Maloha sits quietly between the two speakers.

Study

Education & learning

Practice pronunciation with a native-level partner, read along with parallel transcripts, and pick up phrases mid-conversation.

Personal

Personal connection

Call a grandparent who speaks a different language. Date across a language barrier. Make a friend on a flight.

How it works

Engineered for the gap between words.

Typical end-to-end latency
500ms
you
industry avg ~2s

Measured on LTE from speech-end to translated audio-start. Faster on WiFi.

01

Capture

Local VAD detects speech boundaries and streams audio over WebRTC the instant you start talking.

02

Transcribe

Realtime speech-to-text recognizes words as they land, with confidence scores for post-correction.

03

Translate

A model picked per-pair rewrites each phrase with cultural register preserved — not a word-for-word swap.

04

Speak

Low-latency TTS streams the translated audio back before the next sentence arrives.

Privacy

Your conversations don't stick around.

Maloha processes speech in real time and then forgets it. There's no recording, no transcript database, no training on what you said.

  • No recordings stored on our servers
  • End-to-end encrypted media transport
  • Processed in real time — then discarded
  • Chat history lives locally on your device
  • Open-source components where possible
Get going

Four taps, then talk.

01

Pick two languages

Yours and theirs. Or leave one on auto-detect and let Maloha figure it out.

02

Hand them the phone

Or sit together — Maloha works one-mic or two.

03

Hold to speak

Speak naturally. Maloha catches the end of your sentence and starts translating.

04

Keep the rhythm

They hear it, they reply, you hear theirs. Conversation, not ping-pong.

Stop losing the middle of the sentence.

Maloha is free to try. Download and make a call.

Download on iOSGet it on Android