VoiceTrans+

VoiceTrans+ — Fast, affordable, fully offline AI voice interpreter that learns your words

Just speak, and VoiceTrans+ instantly translates between your two chosen languages both ways and reads the result aloud. The translation AI runs on your device, so normal interpreting is fully offline — no internet needed. Perfect for travel, hospitality, and everyday conversations. It handles long sentences without keeping you waiting, and it gets smarter as you teach it your own terms and phrasing. At about USD 10 per year, it costs a fraction of other apps. No conversation audio ever leaves your phone.

Key features
• Two-way voice translation between two selected languages, fully offline.
• Save sentences you use often as phrases and speak them with one tap. Switch the other language and the phrase is re-translated and spoken in that language.
• Save "packs" of vocabulary and translation style for different scenes, and switch between them.
• Add words to a pack yourself, or have a specified word registered automatically using an online AI.
• Send a conversation to an online AI to refine a pack for more natural translations.
• Online AIs available for these optional features: Gemini, Claude, and OpenAI — each requires your own API key.
• Confidentiality: for work involving personal information or confidential duties, give the pack a name starting with "pro_". While a pro_ pack is in use, the feature that sends conversations to an external AI for learning is disabled.
• Other than these optional AI features — and verifying that your account is valid — the app does not communicate externally.
• UI and translation in 50 languages; device languages that are not supported default to English.
• Some Android devices (Pixel 10 series with Tensor G5, and Snapdragon 8 Elite phones) automatically use a model built for the dedicated AI processor (NPU) and translate faster. Other devices run automatically on the CPU.
• Microphones: uses the built-in mic, or connect a USB mic to use two mics, each set to recognize only one specified language.
• Recognition modes: recognize only while holding the voice button, or auto-translate after a set period of silence (default 1.5 s).
• Print the entire conversation, or print per language.

First launch and startup
• On first launch, a model (about 2.6–3.7 GB — a lightweight E2B to high-accuracy E4B model is auto-selected to match your device; NPU-capable devices use a dedicated model of about 3.1 GB) is downloaded. After the download, initialization (cache build) may take a few minutes; the app is then ready to use. Wi-Fi is strongly recommended.
• On each later launch, loading the model cache takes about 6–12 seconds (varies by device). On slower, older devices, Fast mode (Settings > Fast mode, for older devices) makes translation several times faster; it turns on automatically on devices with little memory. Speaking in chunks of about 3–4 seconds keeps it comfortable.

Pricing
• The first 7 days are free for new users.
• After the free period, a subscription is required:
- General: about USD 10 (United States US$9.99, as of 2 Aug 2026) / year — voice translation in all supported languages (all core features).
- Pro: about USD 40 (United States US$39.99, as of 2 Aug 2026) / year — General plus professional/confidential (pro_) packs.
• Both are annual, auto-renewing subscriptions and include a 7-day free trial.
• Prices shown here are a guide. The actual price varies by country, exchange rate and tax; the price shown on Google Play is the one that applies.

Disclaimer
This app uses AI translation and does not guarantee absolute accuracy. The developer assumes no liability for any damages arising from the use of this app.