Mobile App

Puente

Face-to-face translation that stays on the device.

Hold a natural bilingual conversation offline—speak, hear the translation, keep the turn going—without sending audio to the cloud.

Request a demo Contact sales

Puente offline translator interface showing speech recognition and multiple language pair options

What is Puente?

Puente is an Android conversational interpreter for shared devices. Two people take turns speaking into one device. Puente recognizes speech on-device, translates it, speaks the result aloud, and keeps the transcript—so the exchange stays fast, private, and usable where Wi‑Fi is weak or forbidden after setup.

The problem

Front desks, clinics, hospitality counters, and field teams hit the same wall: a customer or patient who does not share a language, a shared tablet on the counter, and unreliable or restricted connectivity. Consumer translator apps assume personal phones, constant internet, and cloud processing. That breaks privacy expectations, fails in dead zones, and leaves no clean record of what was said.

The solution

Puente is an Android conversational interpreter for shared devices. Two people take turns speaking into one device. Puente recognizes speech on-device, translates it, speaks the result aloud, and keeps the transcript—so the exchange stays fast, private, and usable where Wi‑Fi is weak or forbidden after setup.

Benefits

•
Works offline after setup
First download of translation models, then recognition and translation run without a network.
•
Privacy by design
Speech and translation stay on the device; nothing is sent to a cloud translator for the core flow.
•
Built for turn-taking
Optimized for face-to-face interpreting, not chatbots or typed one-shots.
•
Operational record
Export bilingual conversations as TXT, PDF, or CSV for follow-up and handoff.
•
Shared-device ready
Speaker profiles, push-to-talk, and confidence cues for noisy counters and busy floors.

Features

Conversation modes

Auto English ↔ Spanish with language detection, plus manual English ↔ Spanish, English ↔ French, and English ↔ German. Choose a start language and swap sides when roles flip.

Clear turns under pressure

Push-to-talk, silence auto-stop, mic level feedback, and optional noise reduction keep turns clean in real rooms—not lab silence.

Confidence you can act on

See recognition confidence on each turn. Retry unclear speech or translate anyway when the situation cannot wait.

Transcript and favorites

Searchable conversation history with timestamps. Star important lines so staff can find them later.

Exportable bilingual logs

Export to TXT, PDF, or CSV and share through the system share sheet. Keep an export history on device.

Offline phrasebook

Common English/Spanish phrases for repeated workflows when a full conversation is not needed.

Model control

Model Manager for Vosk speech models and ML Kit translation packs—simple view for day-to-day use, advanced when IT needs visibility.

How it works

1
Set up once
Install Puente, grant microphone access, complete onboarding, and allow the first ML Kit model download.
2
Choose a mode
Auto EN↔ES for open conversation, or lock EN↔ES / EN↔FR / EN↔DE for known pairs.
3
Take turns
Speak (or hold push-to-talk). Puente shows and speaks the translation, then listens for the reply.
4
Keep or share the record
Favorite key turns; export TXT, PDF, or CSV when the visit or desk interaction ends.

Ideal for

•Clinic and reception desks needing bilingual intake without a staff interpreter on every shift
•Hospitality and front-of-house counters on shared tablets
•Field and operations desks where connectivity is intermittent and privacy matters
•Any team that needs offline reliability + exportable bilingual logs, not another consumer translator icon

Trust and privacy

•Speech recognition uses on-device Vosk models (English, Spanish, French, German bundled with the app).
•Translation uses Google ML Kit offline models after the initial download.
•Core conversation flow does not require cloud STT or cloud translation APIs once models are cached.
•Exports leave the device only when a user explicitly shares them.

Supported languages (v1.0)

Platform: Android (API 21+)

Mode	Languages
Auto	English ↔ Spanish
Manual	English ↔ Spanish
Manual	English ↔ French
Manual	English ↔ German

Frequently asked questions

What is Puente?

Puente is an Android conversational interpreter for shared devices. Two people take turns speaking into one device; Puente recognizes speech on-device, translates it, speaks the result aloud, and keeps a bilingual transcript—without sending audio to the cloud for the core flow.

Does Puente work offline?

Yes. After the first download of translation models, speech recognition and translation run on the device without a network. Core conversation does not require cloud STT or cloud translation APIs once models are cached.

Which languages does Puente support?

v1.0 supports Auto English ↔ Spanish, plus manual English ↔ Spanish, English ↔ French, and English ↔ German. Platform: Android (API 21+).

Is speech sent to the cloud?

No. Speech recognition uses on-device Vosk models. Translation uses Google ML Kit offline models after the initial download. Exports leave the device only when a user explicitly shares them.

Can I export conversation logs?

Yes. Export bilingual conversations as TXT, PDF, or CSV through the system share sheet, and keep an export history on device.

Who is Puente for?

Clinic and reception desks, hospitality and front-of-house counters on shared tablets, field and operations desks with intermittent connectivity, and any team that needs offline reliability plus exportable bilingual logs.