Speechmark is a meeting notes tool designed specifically for Mac users, combining recording, transcription, and summarization, with an emphasis on local processing and privacy protection.
Purpose Speechmark is used to fully record meeting content on a Mac, automatically generate structured notes, and help users capture decisions and action items, rather than merely listing bullet points.
Main Capabilities
- Recording: One-click recording from the menu bar, capturing both the microphone and meeting audio, with no browser extension or bot joining the call.
- Speaker separation: Automatically separates speakers. After confirming names for the first time, subsequent transcriptions are labeled with speakers and read like a script.
- Note generation: Automatically generates decisions, action items, and a brief recap, in a plain editorial style. You can choose Apple Intelligence, a local Ollama model, or use your own OpenAI, Anthropic, or Google Gemini key. Click any line to jump to the corresponding moment in the recording.
- Integration with Claude Desktop: One-click connection, installed as a Claude extension, with no configuration file, terminal, or API key required. Claude can answer questions based on local notes and cite which meeting the answer came from. The connector reads notes locally and does not itself make network calls.
How to Use
- Download and run the app, with no account, credit card, or waitlist required.
- Click the record button in the menu bar to start recording.
- After confirming speaker names for the first time, subsequent transcriptions are automatically labeled.
- In settings, choose the model used for summarization (local or cloud model).
- If you want to integrate with Claude, go to Settings → Assistant → Add to Claude Desktop and install it with one click.
Use Cases Suitable for users who value meeting note quality and privacy, such as sprint retrospectives, PRD updates, stakeholder reports, and more. Recordings and transcriptions remain on the Mac, transcription runs locally on Apple silicon, and notes are generated on-device by default.
Limitations and Privacy
- System requirements: macOS 14.2 (Sonoma) or later, Apple silicon Mac.
- Local processing: Audio does not leave the Mac, and transcription runs on-device. If you choose a cloud AI model for summarization (using your own key), only the transcribed text is sent, not the audio. You can also use Apple Foundation Models or a local Ollama model to process everything entirely on-device.- No bot joins the call, and other participants do not see a recording bot.
- No account, login, or registration required. Usage statistics are opt-in and off by default.
- Pricing: One-time purchase, covering up to 3 Macs, with no subscription. You can download it free to try. The founding customer price is $49 (for the first 200 customers), after which it is $79.
Comparison Notes According to official information, Otter and Fireflies are cloud-first, upload audio to servers, and charge per seat; Granola transcribes on-device but syncs notes to the cloud and, unless you opt out, uses anonymous meeting data to improve its AI models. Speechmark keeps audio, transcriptions, and notes on the Mac, requires no account, does not send a bot, and is a one-time purchase.
