Snicksnack – Speech to Text

Snicksnack turns speech into text while you talk. Meetings, interviews, lectures and quick notes are transcribed right on your iPhone. No audio leaves it. Swedish is its specialty, with a speech model trained especially for Swedish. Snicksnack also writes English and 23 other European languages. HOW IT WORKS • Tap the bubble. Recording keeps going in your pocket, even with the screen locked. • Talk. The text grows as you speak. • Tap Stop. The whole recording is transcribed again for better text, and the speakers are sorted out. THE LITTLE EXTRAS • 25 languages: Swedish, English, German, French, Spanish, Italian, Dutch, Polish, Finnish, Danish and more. Pick a second language and switch to it with one swipe. • Who said what: the text is split up by speaker. Name them and the names show up everywhere. • Title, tags and a note on every recording. Dictate them or type. Search everything. • Tap a word to play the audio from there. • Edit right in the text. • Summarize: decisions, to-dos, minutes or quotes. Apple Intelligence does it on the phone, and every point links to its place in the audio. • Quick notes that you merge at the end of the day. • Transcribe audio files from other apps: voice memos, voicemails, interviews. • Apple Watch: record on your watch, the text arrives on your phone. • Shortcuts, the Action button and Control Center. • Share as text, Markdown, subtitles or audio. • Keep the text and delete the audio when you want to save space. • Optional: send a recording to a server you choose, for example at work. Nothing is sent unless you choose to. IT STAYS ON YOUR PHONE The speech models run on your phone. No accounts, no analytics, no ads. You only need the internet the first time, to download the speech model (about 500–650 MB depending on the language). Snicksnack works on iPhone 11 and later with iOS 26. Summaries need an iPhone with Apple Intelligence. Snicksnack is built on Klang Pianissimo for Swedish and Danish, NVIDIA Parakeet for other languages and pyannote community-1 for speakers. All under CC BY 4.0.