Voice Dictation Guide

Google Speech-to-Text in FLAIR: medical dictation and general live dictation for Relevant Findings

Where dictation lives

On the report create page, in-browser voice dictation is only in the AI Assistant panel, under Relevant Findings. There is no separate dictation control for the main Report Content editor — type, paste, or use your usual dictation software there.

Two options (same Google Cloud credentials):
  • Medical dictation (Google) — Google Cloud medical dictation model, tuned for clinical and radiology terminology. You record, then tap stop; transcription runs and text appears in one pass.
  • Live dictation — Google (general) — general Speech-to-Text (not the medical model). Text is sent in short chunks while you speak so words appear progressively.

Medical dictation (Google)

Best when you want maximum help with terms like pneumothorax, cardiomegaly, and hepatosplenomegaly.

  1. Click Medical dictation (Google)
  2. Speak your findings clearly at a normal pace
  3. Click again to stop recording
  4. After processing, transcribed text appears in Relevant Findings

Live dictation — Google (general)

Use for non-clinical wording or when you want text to appear while you are still speaking. Uses the general latest_long model, not medical dictation.

  1. Click Live dictation — Google (general)
  2. Speak; new text is transcribed in chunks and inserted as each chunk returns
  3. Click again to stop
Keyboard: With the cursor in Relevant Findings, F9 starts or stops Medical dictation (Google) (same as clicking that button). It does not start the general live mode.
Voice Commands

You can speak formatting commands while dictating. The Medical dictation (Google) model recognizes these and converts them to punctuation and formatting. General live dictation may handle spoken punctuation differently.

CategorySay this...To get...
Punctuationperiod.
comma,
question mark?
colon:
Line breaksnew line or next lineLine break
next paragraphParagraph break
Listsnext point or next itemBullet point
next numberNumbered item
Sectionsimpression / findingsSection heading
OthercapsCAPITALIZE next word
scratch thatDelete last phrase
Don't stress about getting commands perfect. When dictating into the Relevant Findings box, the AI cleans up all formatting when it generates the report.
Text Insertion

Dictation uses your current cursor or selection in Relevant Findings each time text is applied. For live general dictation, that means you can move the caret or select text between chunks.

Replace

Highlight text → the next inserted text replaces the selection

Insert

Caret in the box → text inserts at the caret

Append

Focus elsewhere → text appends to the end of findings

Tips for Better Accuracy

Speak clearly

  • Normal pace, no need to slow down
  • Pause briefly between sentences
  • Use medical terms confidently

Environment

  • Minimize background noise
  • Mute system notifications
  • Close door if in a shared space
Medical terminology is well-supported: pneumothorax, cardiomegaly, hepatosplenomegaly, lymphadenopathy, atelectasis, and thousands of other radiology terms are recognized accurately.
Troubleshooting

"Could not access microphone"

Grant microphone permissions: click the lock icon in the address bar → Site settings → Allow microphone.

"Transcription failed"

Check internet connection. Try speaking more clearly or closer to the microphone. Ensure audio isn't too quiet.

Incorrect transcription

Reduce background noise. Speak more slowly for complex terms. Review and edit after dictation.

No speech detected

Make sure your microphone is working. Try a different browser if the issue persists.

Need more help?

Contact your IT administrator for technical support.

PowerMic III Setup Back to User Manual