10月7日的Übersetz: Live Translator应用程序分析

Understand spoken conversations as they happen. Übersetz listens through your iPhone or iPad microphone, transcribes speech with WhisperKit, and translates it into your chosen language. Let the app detect the source language automatically or select it yourself. KEEP LISTENING IN THE BACKGROUND Start listening in Übersetz, then switch to another app or lock your device. Microphone capture, transcription, and translation continue until you stop the session or iOS interrupts recording. The system microphone privacy indicator remains visible whenever the microphone is active. Your Heard and Translation text is checkpointed as it changes. If another audio session interrupts listening, or the app closes unexpectedly, reopen Übersetz to recover the latest saved transcript. Only Clear removes the retained session. CHOOSE HOW TO TRANSLATE - Apple Translation for supported language packs downloaded to your device - OpenAI GPT Realtime Translate - Google Gemini Live Translate - Your own Ollama server - OpenAI-compatible endpoints, including OpenRouter SPEAKER DETECTION (BETA) Optional SpeakerKit processing identifies speakers and labels their turns on device. SpeakerKit pauses when Übersetz moves into the background to reduce processing and battery use. Background speech is still transcribed and translated, without speaker labels, and detection resumes when the app returns to the foreground. MADE FOR LIVE READING - Automatic or manually selected source language - Smooth transcripts that follow new text while you stay at the bottom - A large landscape view for following translations across the room - Heard and Translation views - Markdown sharing and file export - Session recovery after interruptions or an unexpected close - Light, dark, and system appearance PRIVACY BY CHOICE WhisperKit transcription runs on your device. With Apple Translation, translation also stays on device after the required language packs are installed. If you choose GPT or Gemini, microphone audio is sent to that provider. Ollama and OpenAI-compatible engines receive transcript text at the endpoint you configure. Credentials are stored in the device Keychain. Übersetz has no ads or tracking. No account is required for on-device use. External translation providers may require their own API key, network connection, and paid account. Requires iOS or iPadOS 26.4 or later. Background recording must be started by you while Übersetz is open. Force-quitting the app stops microphone capture; your most recent completed checkpoint remains available when you reopen it. LATEST IMPROVEMENTS: • Clearer privacy disclosures and consent before microphone use or sending content to external providers. • Better first-run setup for Apple on-device speech, with clearer model status, downloads, and retry options. • Saved provider, language, and appearance choices. • More target languages, script/region variants, available offline. • Clear guidance if a previously selected language is no longer available.