What is verto?

Verto is a native iOS translation app built entirely in SwiftUI, offering three ways to translate: typing text, speaking in a live conversation, and pointing the camera at something like a menu. The README describes it as an experimental testbed for the developer's own translation models and large language model translation engines, though right now translation is powered by the free unofficial Google Translate endpoint. The app targets iOS 17 and above and is fully localized into Simplified Chinese, English, Japanese, Korean, and Spanish. For text translation, a user types or dictates a sentence, taps a confirm button, and the app sends it to Google Translate, caching successful results in memory so repeated translations of the same text do not hit the network again. The result screen lets users swap languages, hear the translation read aloud, copy it, favorite it, or see alternate translations when available. The voice conversation mode is the most developed feature. It listens continuously, shows a live rough translation while the user is still speaking, and automatically detects sentence breaks based on pauses and silence. On iOS 26 and newer it uses Apple's on device SpeechAnalyzer for speech recognition, falling back to the older SFSpeechRecognizer on earlier iOS versions and in the simulator. A notable feature is automatic bilingual detection: the app runs two recognition tracks at once, one per language, and picks whichever one is more confident, so two people can speak different languages back and forth without manually switching. Translation itself prefers Apple's built in Translation framework when available, falling back to the Google Translate endpoint otherwise. The camera mode currently uses local demo data rather than real on device text recognition for the menu translation overlay. The project ships with automated UI tests that use canned, scripted translation and speech responses so the tests never touch the real network, microphone, or text to speech. It can be built and run from Xcode using any iOS 17 or later simulator, though real speech recognition and Apple's on device translation only work correctly on physical devices, since the simulator lacks the on device machine learning hardware they depend on.