🎙️

Voice Assist — Dictate Prescriptions by Voice

Record your clinical notes by voice and get structured prescriptions, lab orders, imaging, and diagnoses in one tap

5 min read

What is Voice Assist?

Voice Assist lets you speak your clinical notes naturally — in Arabic, English, or both — and the system automatically extracts structured prescriptions, lab orders, imaging requests, diagnoses, procedures, and referrals. No typing needed. Just tap the microphone, speak for 30 seconds to 3 minutes, review what the AI extracted, and approve with one tap.

Best for: Post-encounter dictation after the patient leaves. Speak freely about what you want to prescribe, which labs to order, and your diagnosis. The AI understands medical terminology in both Arabic and English.

Step-by-Step Guide

1. Open a Visit

Navigate to an active visit from your dashboard or the visits list. Voice Assist is available on any visit with status Waiting or In Progress. It is not available on completed visits.

2. Tap the Microphone Button

In the visit workspace, you will see three small icon buttons below the patient name. The first button (microphone icon) opens the Voice Assist panel.

Visit workspace showing the microphone button (first icon) below the patient name that opens Voice Assist

3. Grant Microphone Permission

The first time you use Voice Assist, your browser will ask for microphone access. Tap Allow. This permission is remembered for future visits. If you accidentally denied it, see the Troubleshooting section below.

Voice Assist panel showing the Start Recording button, timer, and instructions

4. Record Your Notes

Tap the red Start Recording button. A timer will show how long you have been speaking. Speak naturally — you do not need to use specific commands or formats. Here is an example of what you might say:

"المريض عنده التهاب حلق حاد، هنكتبله أوجمنتين جرام مرتين يومياً لمدة أسبوع، وبانادول إكسترا خمسمية تلات مرات خمس أيام. محتاج صورة دم CBC عاجلة وأشعة صدر. التشخيص J02.9 التهاب البلعوم الحاد. تحويل لأنف وأذن لو مفيش تحسن."

You can mix Arabic and English freely. The AI recognizes drug names in both languages.

5. Stop Recording

Tap Stop when you are done. The system will show "Processing..." while the AI analyzes your audio. This usually takes 3-8 seconds.

6. Review Extracted Items

The Review Panel appears showing everything the AI extracted, organized into sections:

  • Prescriptions — Drug name, dosage, frequency, and duration
  • Lab Orders — Test name and priority
  • Imaging — Type (X-ray, CT, MRI, etc.) and body part
  • Diagnoses — Description in Arabic with ICD-10 code
  • Procedures — Procedure name
  • Referrals — Specialty and reason

Each item has a green checkbox. To exclude an item you do not want, tap the checkbox to deselect it. Excluded items appear greyed out with a "Restore" link.

Review panel showing extracted prescriptions, lab orders, imaging, and diagnoses with approve and exclude controls

7. Approve

Tap the green Approve button. All selected items are saved to the visit in one transaction. You will see the items immediately appear in the visit tabs (Prescriptions, Lab Orders, etc.).

8. Undo (if needed)

Made a mistake? Within 30 seconds after approving, an Undo button appears. Tapping it removes all items that were just created — nothing else on the visit is affected. After 30 seconds, the undo window closes and items must be deleted manually.

Tips for Best Results

  1. Speak clearly — Pronounce drug names distinctly. "أوجمنتين جرام" is better than mumbling.
  2. Include dosage and duration — Say "خمسمية ملجم تلات مرات لمدة خمس أيام" for best extraction.
  3. Mention the ICD-10 code — If you know it, say it. "J zero two point nine" or "جيه صفر اتنين نقطة تسعة". The AI will match it.
  4. Use a quiet environment — Background noise reduces accuracy. The system will warn you if audio quality is low.
  5. Keep it under 3 minutes — The system supports up to 20 minutes, but shorter recordings (30s-3min) give the best accuracy.
  6. One visit at a time — Each recording is for one patient visit. Do not dictate multiple patients in one recording.

What Gets Created

You SaySystem Creates
"أوجمنتين جرام مرتين أسبوع"Prescription: Augmentin 1g, BID, 7 days
"صورة دم عاجلة"Lab Order: CBC, priority: urgent
"أشعة صدر"Imaging: X-ray, chest
"التشخيص التهاب حلق حاد"Diagnosis: Acute pharyngitis (J02.9)
"تحويل لأنف وأذن"Referral: ENT, routine

Limits

  • 10 recordings per day per doctor (resets at midnight)
  • 3 recordings per visit — if you need more, complete the visit and start a new one
  • Maximum audio length: 20 minutes
  • Undo window: 30 seconds after approval

Troubleshooting

Microphone not working

If the browser did not ask for permission or the recording does not start:

  1. Click the lock/info icon in the browser address bar
  2. Find "Microphone" in the permissions list
  3. Change it from "Blocked" to "Allow"
  4. Refresh the page and try again

Recording says "Could not start recording"

This means another app (Zoom, Teams, etc.) is using your microphone. Close the other app and try again.

"Audio quality was low" warning

The AI detected background noise or unclear speech. Items may still be extracted, but review them carefully. Try recording in a quieter environment next time.

AI extracted wrong drug or dosage

Deselect the incorrect item in the review panel and approve the rest. You can then manually add the correct prescription from the Prescriptions tab. Speaking the drug name more clearly next time helps.

Nothing was extracted (all sections empty)

This usually means the audio was too short, too quiet, or contained only non-medical speech. Try recording again with clear medical content.

Supported browsers

Voice Assist works on Chrome, Edge, Safari, and Firefox (latest versions). It works on desktop and mobile. Safari on iOS requires iOS 14.5 or later.

Privacy Note: Audio is processed in real-time and is NOT stored on any server. Only the extracted text and structured data are saved. The original audio recording is discarded after processing.

Was this article helpful?