Record any conversation. Keep every word on your device.

Yes. Synopsule is a single universal app, so one $1.99 purchase on the App Store puts it on both your Mac and your iPhone — and the Apple Watch app comes inside the iPhone app. Add Synopsule Pro and that one subscription works everywhere too.

No — and that follows directly from nothing being uploaded. There's no cloud library and no sync; each device keeps its own recordings. To move a library, use Backup in Settings: Synopsule writes a plain ZIP holding your recordings and, if you include it, the audio. Restore it on the other device and it merges by recording rather than overwriting, so nothing you already had is lost and restoring the same backup twice adds nothing the second time. It's an ordinary ZIP — you can open it with any unzip tool, without Synopsule.

The interface is English only; it isn't localized. Transcription is separate — Whisper handles many spoken languages, and you can set the transcription language in settings.

The iPhone app records through the microphone, so it's made for in-person conversations, interviews, lectures, and voice notes — and it's where the system integrations live: Shortcuts and Siri, the Action button, a Home Screen widget, a Control Center control on iOS 18 or later, and a Live Activity on the Lock Screen and in the Dynamic Island. The Mac app adds system-audio capture (both sides of a call, no bot), the system-wide ⌃⌥⌘R shortcut, meeting-app detection, printing, drag-and-drop, and the built-in MCP server. Everything else matches: on-device Whisper, speaker labels, playback and annotation, search, every export format including PDF, calendar prompts, notifications, and backup.

It records on your wrist for the conversations your phone isn't in your hand for, then hands the recording to your iPhone, which transcribes it. The elapsed time stays legible on the always-on display, a complication shows the running timer and anything still waiting to send, and a transfers list gives each recording its status — including the reason a transfer failed, with retry or delete. Set the speaker count with the Digital Crown and it's remembered for next time. The Watch app is included in the iPhone app and needs watchOS 26.

Yes. Name a voice once and Synopsule builds an on-device voiceprint for that person, then automatically labels them when they turn up in a later recording. Correct a wrong guess and it learns from the correction. A neural voice model is bundled in the app so this works out of the box, offline, on both Mac and iPhone — and the voiceprints never leave your device. You can tune how strict the matching is, and merge or forget any voice.

Nowhere off your device. Audio is captured to a temporary file and processed by Whisper running locally. After the transcript finalizes, the temporary capture file is deleted and a playback-ready copy is kept on the device so you can replay alongside the transcript. Delete it any time. There is no upload, not to Synopsule, not to anyone.

No. Recording, transcription, playback, exports, and search all work without signing in. The only "account" involved is your Apple ID, for the App Store purchase, plus the optional Synopsule Pro subscription if you choose it.

Yes — from the very first launch. A Whisper model ships inside the app, so you can record and transcribe with no download and no connection at all. On a plane, in a basement, or with the cable pulled, Synopsule works the same way it does at your desk. Download a larger model later if you want more accuracy.

No. Capturing system audio is a Mac capability, done through Apple's Core Audio process taps. On iPhone, Synopsule records what the microphone hears, so for remote calls, run it on your Mac.

You still get the part people love — clean, re-readable notes with who said what — but everything runs on the device that recorded it. Cloud notetakers upload your audio to their servers, usually require an account, and often send a bot into your call. Synopsule transcribes locally with Whisper, labels every speaker on-device, keeps the full audio so you can replay any moment (most tools discard it), and saves notes as open Markdown any AI can read instead of locking them in an app. No bot, no account, works offline, and your audio is never uploaded. It's a one-time $1.99 universal app, not a monthly per-seat plan. The honest trade: with nothing uploaded there's no cloud library and no sync between your devices, so moving a library is a backup file you make and restore yourself.

Yes — on Mac. Synopsule ships with a built-in MCP (Model Context Protocol) server. Two clicks in Settings → Connect to AI Apps and Claude Desktop, Codex (app, CLI, IDE), Claude Code, or ChatGPT can list your recordings, search them, and pull a full transcript or summary on demand. The MCP server is the Synopsule app itself, launched with --mcp, so there's nothing extra to download or update. It's local-only and read-only — transcripts stay on the Mac, and audio is never exposed. On iPhone, the same Markdown exports show up in Files for sharing into the ChatGPT or Claude iOS apps.

Yes. Before you summarize, pick a template — general, one-on-one, standup, sales call, interview, lecture, clinical SOAP, legal, or user research — and Synopsule produces a structured, speaker-aware recap tuned to that format, with action items and follow-ups. No template is behind the subscription; what differs is the engine, since the on-device one returns a general summary whichever you pick. Switching templates keeps the recap you already had, and re-running a summary leaves the action items you've ticked off ticked. You can also keep time-anchored notes as you record or read: tap a note to seek straight to that moment, and your notes fuse into the summary as authoritative context.

Whichever you prefer. Once a summary provider is set up you can switch on automatic summaries, and a recording is summarized the moment its analysis finishes. Leave it off — which is where it starts — and nothing is summarized until you press the button. And if you press Summarize before a provider is set up, Synopsule remembers which recording you meant and runs it the moment setup finishes; back out of setup and nothing runs.

Never the audio. AI is opt-in, and that's still true with automatic summaries on — you've chosen the destination once instead of every time, and a consent sheet names it before any text is sent. Bring your own OpenAI, Anthropic, or Gemini key and transcript text goes from the app straight to that provider. Or keep it on the machine: a local model through Ollama on a Mac, or Apple Intelligence on eligible hardware. Or subscribe to Synopsule Pro and summaries route through Synopsule to OpenAI. Either way, only transcript text ever leaves, never the audio. If you never summarize, nothing leaves at all.

Inside the transcript. Search matches the words people actually said across every recording, with speaker: filters, and searching by meaning covers the transcript body too — so a meeting you never summarized is still findable. That's also what lets Ask your meetings quote lines the transcript really contains and open at the moment they were said. The index is built on your device and rebuilds itself from your recordings if it's ever lost.

If you allow notifications, yes. Analysis of a long recording takes minutes, and the point is that you can walk away — so Synopsule posts a notification when the transcript is ready, when a recording from your Apple Watch has landed, and when analysis stopped without a transcript so you know to act. They're local notifications from the app on your device; nothing is sent to a server to deliver them.

If you let it read your calendar. Synopsule reads it on the device — read-only, never written to, nothing stored or sent — and offers to record when a meeting is starting, naming the recording after the event. On a Mac it can also notice when a meeting app opens and offer then. Both are prompts you accept, not automatic recording, and both stay off until you switch them on.

Yes, on iPhone. Start Recording and Stop Recording are App Intents, so they work in Shortcuts, from Siri ("Start recording with Synopsule"), from the Action button, from a Home Screen widget, and from a Control Center control on iOS 18 or later. While a recording runs, a Live Activity keeps the elapsed time on the Lock Screen and in the Dynamic Island, with stop and pause on it. On a Mac, ⌃⌥⌘R starts and stops from any app.

Yes. Import an M4A, MP3, WAV, or other file the device can decode and Synopsule treats it exactly like something it recorded itself — speaker labels, playback, search, exports, and an optional summary. It's processed on the device like everything else.

Version 2.0 was built for it. Text scales with Dynamic Type on iPhone and Apple Watch, where every font used to be a fixed size. The eight speaker colours were re-derived so they stay apart under simulated protanopia, deuteranopia, and tritanopia, and each speaker also carries a distinct shape when Differentiate Without Color is on — colour is never the only signal. Text holds AA contrast, Reduce Motion is honoured on all three apps, touch targets are at least 44pt, and VoiceOver gets real state, traits, and actions rather than bare labels.

Yes — two on-device paths. On a Mac, install Ollama, pull a model, and point Synopsule at it in AI setup; summaries are then generated entirely on your Mac, with no key or subscription. And on Macs and iPhones with Apple Intelligence (macOS 26 / iOS 26 on eligible hardware), Synopsule can summarize using Apple's on-device Foundation Models — again, nothing leaves the device. Either way, no transcript text and no audio is uploaded. (Ollama is Mac-only; if your iPhone isn't eligible for Apple Intelligence, use a BYO key or Synopsule Pro.)

A Mac on macOS 15 Sequoia or later, or an iPhone on iOS 17 or later, optimized for Apple silicon. The Apple Watch app is included in the iPhone app and needs watchOS 26. Pick a smaller Whisper model for snappy transcription on any supported device; on Apple silicon the larger models still feel close to real time. A model download shows real progress, can be cancelled, and checks you have the space before it starts.

Word (.docx), PDF, Markdown, plain text, interactive HTML, and SRT and VTT subtitles — all of them on both Mac and iPhone. Synopsule can also send a transcript straight into Obsidian or Apple Notes in one tap, and save the recorded audio itself. Speaker labels, timestamps, your notes, your flagged moments, and the AI summary (if you made one) travel with the file, and you pick what to include.

A $1.99 one-time purchase on the App Store, covering Mac, iPhone, and the Apple Watch app. Everything that makes the app what it is — recording, on-device transcription, speakers, playback, annotation, search, every export format, backup — is in that price, with no subscription. Summaries are the only paid-optional part, and three ways to run them cost nothing: your own API key, a local model via Ollama on a Mac, or Apple Intelligence on eligible hardware. Synopsule Pro ($4.99/month or $39.99/year, every device) is for when those aren't enough: an hour-long meeting summarized in one pass instead of in chunks, all nine templates returning their own sections, Ask answering from long excerpts with citations, and no API key to manage. Its trade-off, plainly: transcript text goes to the Synopsule service and on to OpenAI.