Otter.ai
AI meeting assistant for transcription and automated note-taking
Say
Browser‑based AI transcription app turning speech into editable notes
Say lets you record, transcribe, and edit spoken notes directly in the browser using Whisper, T5 and Transformers.js, with real‑time waveforms and local storage.
Overview
Overview
Say is a client‑side voice transcription tool aimed at developers, students, and professionals who need quick, private note‑taking without relying on external services. By leveraging Whisper, T5 and Transformers.js, the app runs entirely in the browser, keeping data on the user’s device.
Core Capabilities
The interface provides intuitive browser‑based recording, real‑time waveform visualization, and AI‑powered speech‑to‑text conversion. Transcriptions appear in a rich‑text editor powered by TinyMCE, allowing formatting, editing, and immediate refinement. All notes are persisted in local storage, ensuring privacy and offline access. The UI, built with React, TypeScript, Tailwind CSS, and Vite, delivers a responsive experience across modern browsers.
Deployment
Installation follows standard Node.js workflows: clone, npm install, and npm run dev for local development. Production builds are generated with npm run build, outputting static assets ready for any static‑host (e.g., GitHub Pages, Netlify). No server‑side components are required, making deployment straightforward and cost‑free.
Highlights
Browser‑based recording with real‑time waveform visualization
In‑browser AI transcription using Whisper, T5 and Transformers.js
Rich text editor for immediate formatting and correction
Local storage of notes for privacy and offline access
Pros
- No server or cloud dependency; runs entirely in the browser
- User data stays local, enhancing privacy
- Fast development and production builds with Vite
- Modern, responsive UI built with Tailwind CSS
Considerations
- Requires a modern browser with WebAssembly support
- Heavy models may tax low‑end devices
- No built‑in cross‑device synchronization
- Limited to relatively short audio segments for optimal performance
Managed products teams compare with
When teams consider Say, these hosted platforms usually appear on the same shortlist.
Looking for a hosted option? These are the services engineering teams benchmark against before choosing open source.
Fit guide
Great for
- Developers prototyping voice‑driven note‑taking features
- Students capturing lecture snippets on the fly
- Professionals needing quick, private voice memos
- Privacy‑conscious users who prefer local data storage
Not ideal when
- Users needing automatic cloud backup and sync across devices
- Large‑scale transcription of long recordings
- Environments without reliable microphone access
- Outdated browsers lacking WebAssembly support
How teams use it
Meeting minutes capture
Record discussions, get instant transcriptions, edit, and export formatted notes for team reference.
Lecture note taking
Capture key lecture segments, transform them into editable text, and integrate with study materials.
Idea brainstorming
Speak ideas aloud, refine them instantly in the rich‑text editor, and keep them stored locally.
Customer support logs
Agents dictate call summaries, produce searchable notes instantly, and retain them securely on their device.
Tech snapshot
TypeScript92%
JavaScript6%
HTML2%
CSS1%
Frequently asked questions
Do I need an internet connection to use Say?
All processing runs locally in the browser; only the initial download of model files requires a network connection.
Where are my transcribed notes saved?
Notes are stored in the browser's local storage and remain on the same device unless manually exported.
Which browsers are supported?
Any modern browser with WebAssembly support, such as recent versions of Chrome, Edge, Firefox, and Safari.
Can I export my transcriptions?
You can copy or download the edited content from the rich‑text editor; the app does not provide dedicated export formats.
How can I improve transcription accuracy?
Use a clear microphone, speak clearly, and ensure your device has sufficient resources; the underlying models run locally.
Project at a glance
Active- Stars
- 369
- Watchers
- 369
- Forks
- 39
LicenseMIT
Repo age1 year old
Last commit2 months ago
Primary languageTypeScript
Last synced 3 hours ago