1 Project vision

• VocalNoteOne lets anyone dictate voice notes on their computer — no complexity, no subscription, no account to create.
• The application is designed to get to the point: speak, save, reread, share.
• It suits busy professionals just as much as people for whom writing is a daily challenge.
• VocalNoteOne is a tool, not a service.

2 Concrete use cases

• Answering an email quickly. You receive a message, read it, dictate your reply, click “Copy” and send without touching the keyboard.
• Notes on the fly. An idea, a task or a reminder is dictated, saved immediately, and stays available and editable at any time.
• Creative writing and drafting. Dictate ideas, chapters or dialogue as inspiration strikes, then export the lot as a text file. Suited to authors, journalists and students.
• Accessibility. A simple, dignified alternative for people with dyslexia, dysphasia or dyspraxia, people with limited literacy, or anyone for whom writing is an obstacle.
• Talking to an artificial intelligence (an LLM). On a phone, voice dictation is built into the keyboard. On a computer, that feature generally doesn't exist. VocalNoteOne fills the gap: dictate your question or instruction, paste the text into the conversation, and the exchange goes considerably faster than typing — a use particularly suited to anyone working with this kind of tool day to day.

3 Features

Dictation

• Real-time transcription, editable on the fly while you dictate.
• Immediate correction from the keyboard without breaking the flow; manual corrections are preserved.

Status and instruction separated across two channels

The banner at the top, above the transcript area, carries only the state of dictation. A dedicated line just above the buttons — always visible and announced to screen readers — carries any instruction requiring action from the user.

Network diagnosis before microphone diagnosis

When offline, the browser sometimes returns a permission error even though the problem lies elsewhere. The application now checks network state first, so it doesn't send the user off to a microphone setting when the real cause is a dropped connection.

• Resumption after prolonged silence. Automatic restart with a capped, increasing delay, a clean stop, and an explicit message if the service stops responding.

Notes

• Automatic saving when dictation stops, with a full dated history.
• Editing by clicking directly on a note's text, with Save, Cancel, Ctrl + Enter, and confirmation before any edit is discarded.
• Deletion with confirmation.
• One-click copying, with an automatic fallback if the browser restricts clipboard access.
• Full export as .txt: all notes with their dates in a single timestamped file.
• Automatic draft of the text in progress, to prevent any loss if the file is closed accidentally.
• Transactional saving. If local storage fills up, the note stays on screen and is kept as a draft rather than lost. The counter cannot diverge from what is actually written to disk.
• Separation by language. Each language version has its own storage space, with no collision between them.
• Already-saved notes accessible offline, since they live on the user's computer; only dictation itself requires a connection.

Comfort

• Built-in zoom, no setting required.
• Dark mode. Preference remembered and applied before the first paint, with no flicker on load.
• 8 languages available: French, English, Spanish, Brazilian Portuguese, Italian, German, Romanian, Dutch. Each language ships as a dedicated application.

4 Accessibility and design

• An interface built for clarity above all, with a clean design.
• High-contrast colours (orange, blue, white) and legible typography, suited to people with low vision and colour blindness.
• All interface text meets the WCAG 2.1 level AA contrast threshold, including the in-progress dictation text, the instruction line and secondary labels.
• No unlabelled controls. Every action on a note carries its name in writing next to the icon. An icon alone assumes a decoding effort the intended audience shouldn't have to make.
• Simplified use: two main buttons and one large text area.
• Full keyboard navigation, with a visible focus indicator on every interactive element.
• Screen reader compatibility: state changes are announced automatically, including the instruction line at the bottom of the screen; the notes window is declared as a dialog, with focus return handled. Closing and cancelling via the Esc key.
• Animations are disabled automatically if the user's system requests it.

5 Supported browsers and first use

VocalNoteOne is built on the Web Speech API, whose availability and quality vary by browser.

Caveats regarding Microsoft Edge

Two points to know before any deployment in a professional environment.

1. The recognition engine is different. Edge does not use Google's service but Azure Cognitive Services, a Microsoft service. Detailed implications in section 6.
2. Recognition can be disabled by enterprise policy. Microsoft exposes an administrative policy, SpeechRecognitionEnabled, deployable by GPO or registry key, which cuts off access to the Web Speech API. When disabled, the application loads normally but no text is produced. Any large-scale Edge deployment must be validated with the IT department.

In practice, continuous dictation is steadier in Chrome, which remains the browser recommended to users.

Steps for use

1. Open the file VocalNoteOne.html with Google Chrome.
2. Click the blue 🎙️ Record my notes button.
3. In the window that opens, click Allow while visiting the site.
4. Speak: your text is written in real time.

Microphone permission when running locally

A file opened directly from disk is served over the file:// protocol, which gives the browser no stable origin to attach a persistent permission to. Chromium browsers therefore cannot remember microphone access permission and ask for it again.

Observed and measured behaviour. After roughly 20 seconds without speech, Chrome interrupts the recognition session. The application restarts it automatically, which triggers a fresh permission request. The options “Allow while visiting the site” and “Allow this time” behave identically in this context: the request comes back either way.

The application's response. The request cannot be removed by code — it stems from the browser's security model. The application therefore makes it explicit on the dedicated channel described in section 3, which stays visible even when Chrome's window covers the top of the screen. The text already dictated stays on screen and dictation resumes in the same place. This state is also announced to screen readers.

Removing the constraint entirely. Serving the application from a stable origin — a local server on http://localhost, or a future packaged version — lets the browser record the permission permanently. That is the configuration to favour for a workstation deployment or intensive use.

6 Privacy and commitment

• No collection of personal data by VocalNoteOne.
• No trackers, no advertising, no account required.
• Notes are stored solely on the user's computer, in the browser's local storage, and remain exportable at any time.
• VocalNoteOne neither retains, transmits nor processes any voice data.

Model and support

Free, no account, no advertising, no data collection. The project lives solely on donations from those who choose to give. The addresses appear at the bottom of the application, on four networks — Bitcoin, Solana, Ethereum, Arbitrum — and grant no additional functionality.

Voice processing by the browser

Speech recognition is not performed by VocalNoteOne but by the browser, which transmits the audio stream to its own service:

• Under Google Chrome: processing carried out by Google LLC's servers.
• Under Microsoft Edge: processing carried out by Azure Cognitive Services, a Microsoft Corporation service. Microsoft explicitly documents that voice data leaves the machine.

The data controller therefore differs according to the browser chosen. Any organisation subject to the GDPR must account for this in its record of processing activities and determine the authorised browser accordingly. VocalNoteOne plays no part in either of these flows and has no access to them.

7 Availability

• HTML version: to be opened in Google Chrome or Microsoft Edge. Compatible with Windows, macOS and Linux. No installation, no dependencies, a single file.

A packaged version for Windows is on the roadmap. It is not available at this time.

Contact and user feedback: contact.vocalnoteone@gmail.com

8 Legal notices

• Application provided “as is”, without guarantee of permanent or error-free operation.
• The user alone is responsible for the use made of the application.
• Any abusive, illegal or improper use is strictly prohibited.
• Speech recognition is provided by the browser's Web Speech API: by Google LLC under Google Chrome, by Microsoft Corporation (Azure Cognitive Services) under Microsoft Edge. VocalNoteOne is affiliated with neither Google nor Microsoft.