Data Processing & Patient Notification Guidance

1. Introduction

HearScribe processes voice data differently from traditional recording systems. This document explains our unique architecture and helps clinicians understand their compliance position regarding patient notification.

Key Insight

HearScribe's architecture means no patient audio is ever stored on HearScribe servers. Voice data follows two parallel paths: local browser recording (never transmitted) and live transcription by your web browser's built-in speech service.

2. How HearScribe Processes Voice Data

When you use HearScribe's transcription feature, your voice data is processed through two parallel systems:

Data Flow Overview

Microphone Input

↓

Browser (Your Device)

↓ Splits into two parallel paths:

Path 1: Local Recording

Audio → MediaRecorder API → Browser Memory (RAM)

Never sent to any server. Destroyed on tab close or "Next Patient".

Path 2: Live Transcription

Audio → Web Speech API → Your browser's speech recognition service (Google for Chrome, Microsoft for Edge)

Processed by the browser provider under its own terms. HearScribe never receives or stores the audio.

Text returned → Your Browser

2.1 Local Recording (MediaRecorder API)

Storage location: Browser memory (RAM) only
Transmission: Audio is NEVER sent to any server
Persistence: Destroyed when browser tab closes or "Next Patient" is clicked
Download option: User can choose to download the recording to their device

2.2 Live Transcription (Web Speech API)

Processing: Audio is streamed to your browser provider's speech service (Google for Chrome, Microsoft for Edge) for real-time text conversion
Storage: HearScribe never receives or stores the audio; the browser provider processes it under its own terms

2.3 Note Generation (Google Gemini API)

Input: Transcribed text, any patient details you entered and any clinical images are sent to the Google Gemini API
Processing: HearScribe retains nothing after generation; Google processes the request under its Gemini API terms
Output: AI-generated notes returned to your browser only

2.4 Patient Identifiable Data

Patient names and other details you enter are never stored on HearScribe's servers. They are included in the text sent to Google Gemini to generate your notes, and in emails you choose to send (see Email Delivery below), and are then discarded.

3. Two Operating Modes

HearScribe offers two modes that affect how audio recordings are handled after you stop recording:

Mode A: Audio Player Visible (Default)

After stopping recording, audio playback controls and download button are available.

The recording exists in browser memory until:

User clicks "Next Patient" (clears all data)
User closes the browser tab
User downloads the file (then it's on their device)

Patient could potentially see or hear the playback during the consultation
Recording can be downloaded and retained

Recommendation: Inform patients that a recording is being made, as they could potentially access it during the consultation.

Mode B: Audio Player Hidden (Privacy Mode)

Audio playback controls and download button are not displayed after recording.

Recording still exists in browser memory during session but is inaccessible
Clicking "Next Patient" or closing the tab destroys all audio data
No persistent recording is ever created
No way to play back or download the audio

Recommendation: Patient notification is at the clinician's discretion and professional guidance. No accessible recording exists, though live audio is still processed by the browser's speech service.

You can switch between these modes in your Account Settings under Privacy Settings.

4. Third-Party Data Processing

Live transcription uses the speech recognition built into your web browser (the Web Speech API). In Chrome, audio is processed by Google; in Edge, by Microsoft. This is the browser provider's own service, governed by its terms - it is not a service HearScribe contracts or controls, and it is not the same as Google Cloud Speech-to-Text.

4.1 Email Delivery (optional)

Email Notes Feature

When you use the Email Notes feature, documentation is sent via ZeptoMail (Zoho's transactional email service, EU data centres) over an encrypted TLS connection. HearScribe stores no copy of the email. If a message cannot be delivered, the sender receives an automated non-delivery notification. This is encrypted in transit but is not an NHS-accredited secure email service.

Message content: Content storage is disabled on our ZeptoMail account - the body of your notes emails is not retained after delivery
Delivery logs: Sender, recipient, subject line and delivery status are retained by ZeptoMail for 60 days for delivery troubleshooting, then deleted. Note that subject lines include the patient name you entered
Safeguards: GDPR-compliant service holding ISO/IEC 27001, ISO/IEC 27701 and SOC 2 Type II certifications, covered by Zoho's Data Processing Addendum
Account emails: HearScribe account emails (confirmations, password resets and team invitations) are also delivered via ZeptoMail; these contain no patient information

If you prefer not to use the email feature, you can copy your notes and send them through your own systems. For more details, see Zoho's ZeptoMail GDPR documentation.

5. Patient Notification Guidance

5.1 When Notification is Recommended

Audio player is visible (recording can be played back or downloaded)
You intend to keep a copy of the recording
Your professional body or employer requires it
Local regulations mandate disclosure of any recording

5.2 When Notification May Not Be Required

Audio player is hidden (Privacy Mode enabled)
No recording will be retained
You are only using transcription to assist with note-taking

Important Caveat

This guidance reflects our understanding of the technical data flows. HearScribe cannot provide legal advice. Clinicians should consult their professional body guidelines and, where necessary, legal counsel for definitive compliance requirements in their jurisdiction.

6. Compliance Framework Positioning

HearScribe's browser-only architecture provides a strong foundation for compliance across multiple jurisdictions:

Regulation	HearScribe Position
UK GDPR	No patient data stored on HearScribe's servers. You remain the data controller; HearScribe acts as your processor when passing data to the AI and email services (see our Data Processing Agreement, available from contact@hearscribe.com).
HIPAA (USA)	No PHI stored by HearScribe. Browser speech recognition and the Gemini API used here are not covered by a Business Associate Agreement, so HearScribe should not be treated as HIPAA-compliant for PHI.
Privacy Act (Australia)	Browser-only processing minimises privacy impact.
PIPEDA (Canada)	Similar browser-only architecture benefits. Personal health information remains under your control.

7. Frequently Asked Questions

Does HearScribe store my patient's audio recordings?

No. Audio recordings exist only in your browser's memory and are never transmitted to HearScribe servers. They are automatically destroyed when you close the browser tab or click "Next Patient".

What happens to the audio used for live transcription?

It is streamed to your browser's speech service (Google for Chrome, Microsoft for Edge) for real-time text conversion, under that provider's terms. HearScribe never receives or stores it.

Can I use HearScribe without notifying patients?

This depends on your jurisdiction and professional requirements. With Privacy Mode enabled (audio player hidden), no accessible recording exists, making the process more analogous to traditional note-taking. However, consult your professional body guidelines for definitive guidance.

What's the difference between Audio Player Visible and Hidden modes?

In Visible mode (default), you can play back and download recordings after stopping. In Hidden mode (Privacy Mode), the audio player and download options are not displayed, meaning no accessible recording exists - it's destroyed when you move to the next patient or close the tab.

Is HearScribe HIPAA compliant?

HearScribe stores no PHI, but the browser speech service and AI service it uses are not covered by a Business Associate Agreement, so it should not be used as a HIPAA-compliant service for PHI. Check with your organisation's compliance lead.

Where can I enable Privacy Mode?

Go to Account Settings and look for the "Privacy Settings" section. Toggle "Hide Audio Player" to enable Privacy Mode.

Does Privacy Mode affect transcription quality?

No. Privacy Mode only affects whether the audio player is displayed after recording. The transcription process works identically in both modes.

← Back to Home