Capture without waiting • Save an idea the moment you stop speaking. Transcription happens in the background and the text fills in by itself. • If a recording can't be transcribed, the idea says so and you can try again. • Silent stretches no longer turn into captions like [Music]. More you can do with your ideas • Edit a saved idea to fix a misheard word — it stays in its place in the list. • Share an idea to any app. • Export all your ideas to a single text file you can keep anywhere. • Delete all ideas at once from Settings. Voice recordings, if you want them • Turn on Keep recordings to save the original audio with each idea and play it back. • Settings shows how much space recordings use, and lets you delete them. Speech models • Smaller downloads, a new Turbo model, and dedicated English-only models for English. • Downloads resume after an interruption, keep going when you leave the screen, and check for free space first. • Models are no longer included in iCloud backups. Privacy • Speech recognition now runs entirely on your device. The OpenAI Whisper API option has been removed. Plus stability fixes and an updated privacy policy.

New logic for speech to text feature.