Live Transcriber: Offline AI

LiveTranscriber 1.2 is a major update: • A new Apple Watch app lets you start, pause, resume, and stop recordings from your wrist, play recordings on the watch, and transfer them to your iPhone • New Apple Watch widgets and complications provide quick recording access and show recording and transfer status • Choose Voice, Balanced, or High Quality recording on Apple Watch, with optional recording location • Added Parakeet V3 for fast, high-accuracy Core ML transcription across supported European languages • Start offline background retranscription from the recording player, with progress preserved for long-running tasks • Added optional Sortformer live speaker diarization with stable speaker labels during recording • Added Qwen3-ASR Live Beta and improved streaming support for long recordings • Reworked Nemotron 3.5 Core ML live transcription with improved model loading, background handling, and timestamp alignment • Added optional automatic MOSS post-recording refinement for more accurate final transcripts and speaker-separated results • Added Picture in Picture captions so live transcription remains visible while using other apps • Improved live and saved-transcript translation, with a new settings page for checking and preparing Apple Translation language packs • Added Standard and Spatial Audio capture profiles, sample-rate and mono/stereo controls, wind-noise removal, high-quality Bluetooth recording, and microphone selection • Redesigned recording details and playback with clearer waveforms, transcript locating and following, audio effects, and spatial-audio master sharing • Improved recording editing for titles, key points, tags, summaries, languages, categories, and locations, including location search and current-location support • Added automatic recording details and intelligence output-language controls, with improved local summaries, recording chat, and translated-transcript analysis • Choose the app’s interface language independently from the system: English, Simplified Chinese, Traditional Chinese, Japanese, German, French, or Dutch • Language menus now remember frequently selected languages and prioritize them across the app • Added a redesigned step-by-step introduction covering languages, recording formats, local models, and privacy • Improved model download progress, deletion, and status reporting, with a complete third-party model and runtime license catalog • Fixed the Parakeet model permission error, Nemotron timestamp drift, and multiple background transcription, system-audio, and recording stability issues