Voice Transcription

Praxi's voice transcription converts your session conversations into clinical notes, saving documentation time while maintaining quality.

Overview

Voice transcription uses AI-powered speech recognition optimised for healthcare to convert session audio into text in real time. It provides:

Feature	Description
Real-time capture	Transcribe as you speak with streaming results
Speaker identification	Distinguish practitioner from client
Medical terminology	Accurate healthcare vocabulary (DSM-5, medications)
Privacy-first	Compliant with Australian Privacy Act 1988
Note generation	Convert transcript to structured notes
Live clinical guidance	Real-time suggestions during the session
Session markers	Automatic detection of topic changes and key moments

Australian Healthcare Compliance All voice transcription processing

follows Australian Privacy Principles (APPs) under the Privacy Act 1988. Audio data is processed on Australian-hosted infrastructure, never stored offshore, and you maintain full control over retention and deletion. :::

Getting Started

Hardware Requirements

For optimal transcription:

Equipment	Recommendation
Microphone	USB headset or desktop mic
Audio quality	Clear capture of both speakers
Internet	Stable connection (5+ Mbps)

Recommended microphones:

• Jabra Evolve2 (headset)
• Blue Yeti (desktop)
• Rode NT-USB Mini (compact)

Initial Setup

Follow these steps to enable voice transcription:

1. Navigate to Settings > Clinical > Voice Transcription
2. Click the Enable Voice Transcription toggle
3. Select your preferred microphone from the dropdown
4. Click Test Audio to verify your setup
5. Speak a test phrase and confirm the waveform appears
6. Adjust sensitivity if needed (slider below the waveform)
7. Click Save Settings to confirm

Consent Configuration

Client consent is required for transcription under Australian privacy law:

1. Navigate to Settings > Clinical > Voice Transcription > Consent
2. Enable Require consent prompt before sessions
3. Customise the consent text to match your practice's privacy policy
4. Configure Signed consent storage (digital or paper-based)
5. Set Opt-out handling behaviour (pause session or proceed without transcription)
6. Enable Consent reminder for returning clients if desired

Using Voice Transcription

Starting a Session

1. Open the appointment
2. Click Start Session
3. Voice transcription begins automatically (if enabled)
4. Microphone icon shows recording status

Manual Start/Stop

If not using auto-start:

1. Click the Microphone icon
2. Transcription begins
3. Click again to pause/stop
4. Resume as needed

During the Session

The transcription panel shows:

• Live transcript text
• Speaker labels (Practitioner/Client)
• Confidence indicators
• Timestamp markers

Session Controls

Control	Action
Pause	Temporarily stop transcription
Resume	Continue transcription
Mark	Add a timestamp marker
Delete segment	Remove confidential content

Transcription Modes

Full Session Mode

Transcribes the entire session:

• Continuous streaming recording
• Complete conversation captured
• Session markers extracted automatically (topic changes, interventions, disclosures)
• Best for post-session note generation

Dictation Mode

For practitioner-only dictation:

• Only your voice captured
• Direct-to-notes format
• Supports voice commands: "new paragraph", "new heading", "period", "comma", "delete last", "undo"
• Ideal for post-session documentation

Multi-Speaker Mode

For couples and group sessions (Pro plan and above):

• Identifies and labels multiple speakers
• Speaker enrolment for improved accuracy across sessions (voiceprint-based)
• De-identified notes generated per participant
• Requires separate consent from each participant

Privacy and Security

Australian Privacy Act Compliance

Voice transcription is designed to meet Australian healthcare privacy requirements:

Requirement	How Praxi Complies
APP 1 - Open management	Clear privacy policy, transparent AI use
APP 6 - Use and disclosure	Audio used only for clinical documentation
APP 11 - Data security	AES-256 encryption, Australian data centres
APP 12 - Access	Clients can request transcripts of their sessions

Data Handling

Voice data security:

Stage	Protection
Capture	TLS 1.3 encrypted in transit
Processing	Australian-hosted servers, no offshore transfer
Storage	AES-256 encrypted at rest
Audio retention	Deleted when the linked clinical note is signed; 72-hour safety net regardless of note status
Transcript	Redacted 7 days after the clinical note is signed; only metadata retained
Clinical notes	7-year retention (AHPRA requirement)

Consent Requirements

Under Australian healthcare privacy requirements, before transcription you must:

1. Inform the client that the session will be transcribed
2. Explain the purpose (clinical documentation)
3. Document consent in the client record
4. Respect opt-out requests immediately
5. Provide information about how to access or delete transcripts

Consent is valid for 12 months by default. Praxi automatically prompts you to renew consent when it approaches expiry. Couples sessions require a separate consent model selection (no secrets, secrets, or case-by-case).

Client Consent Required Always obtain explicit, informed consent

before using voice transcription. Under the Privacy Act 1988, clients have the right to know how their information is collected and used. Document consent in the client record before each session. :::

What's NOT Recorded

Transcription automatically excludes:

• Pre-session waiting room time
• Payment and billing discussions (configurable)
• Breaks during session
• Post-session administrative discussion

Data Retention and Deletion

In accordance with AHPRA record-keeping requirements:

• Default retention: 7 years from last contact (adult clients)
• Minor clients: Until client turns 25, then 7 years
• Manual deletion: Available with audit trail
• Automatic purge: Configurable after retention period

Clients may request access to or deletion of their voice transcripts at any time, subject to clinical record-keeping obligations.

Live Clinical Guidance

During active telehealth and phone sessions (Pro plan and above), the AI can show live reflection prompts based on the running transcript. In-person, couples and group sessions do not show live guidance.

Guidance Type	Description
Reflective questions	Follow-up questions grounded in what was said
Goal check-ins	Reminders about goals in your documented plan
Homework review	Prompts to follow up on previously assigned homework

You can ask Praxi a question at any time from the box at the bottom of the guidance panel.

Live guidance does not assess risk or suggest techniques, interventions or assessments. No automated risk alerts are raised.

Live clinical guidance is a support tool -- it runs alongside your

session without interrupting. You can dismiss suggestions you do not need, or add them directly to your session notes. :::

Converting to Notes

Automatic Note Generation

After the session:

1. Review the transcript
2. Click Generate Note
3. Choose your note template (SOAP, DAP, BIRP, progress note, or custom)
4. AI creates a structured note with sections pre-filled from the transcript
5. Review and edit the generated content
6. Add your clinical observations and risk assessment
7. Sign the note

Note Templates

Generated notes follow your preferred template. The default template is automatically chosen based on your profession (for example, dietitians default to ADIME, social workers to DAP).

• SOAP -- Subjective, Objective, Assessment, Plan
• DAP -- Data, Assessment, Plan
• BIRP -- Behaviour, Intervention, Response, Plan
• Narrative -- Free-form clinical narrative
• ADIME -- Assessment, Diagnosis, Intervention, Monitoring, Evaluation (dietetics)
• ICF -- International Classification of Functioning framework
• Functional -- Functional outcomes-focused documentation

What AI Extracts from Transcription

The AI automatically identifies and structures:

• Presenting concerns discussed
• Interventions used and client responses
• Clinically significant direct quotes (preserved verbatim)
• Homework and tasks agreed
• Risk indicators
• Topic changes and session flow

Editing Generated Notes

Always review AI-generated notes:

1. Check accuracy against your clinical memory
2. Add clinical observations (non-verbal cues, body language)
3. Include your professional assessment
4. Verify any direct quotes are accurate
5. Review flagged items marked with [VERIFY] for accuracy
6. Add or update the risk assessment

Speaker Identification

How It Works

The AI identifies speakers by:

• Voice characteristics
• Speaking patterns
• Conversation context

Improving Accuracy

For better speaker identification:

1. Speak clearly at session start
2. Avoid overlapping speech
3. Position microphone between speakers
4. Use consistent seating arrangement

Manual Correction

If speaker labels are wrong:

1. Click the transcript segment
2. Change speaker label
3. AI learns from corrections

Quality and Accuracy

Accuracy Factors

Factor	Impact
Audio quality	High impact
Background noise	Reduces accuracy
Accents	Generally handled well
Medical terms	Optimised for healthcare
Mumbling	May miss content

Improving Quality

1. Position microphone - Equal distance from both speakers
2. Reduce noise - Close doors, minimise interruptions
3. Speak clearly - Especially for key clinical terms
4. Review transcripts - Regular review improves AI learning

Medical Terminology

Praxi's transcription is trained for:

• DSM-5 diagnostic terms
• Medication names
• Clinical abbreviations
• Australian healthcare terminology

Troubleshooting

No Audio Detected

1. Check microphone connection
2. Verify browser permissions
3. Test in system settings
4. Restart browser

Poor Transcription Quality

1. Move closer to microphone
2. Reduce background noise
3. Check internet connection
4. Try different microphone

Speaker Confusion

1. Increase speaker separation
2. Avoid overlapping
3. Speak naturally (not too fast)
4. Manually correct labels

Subscription Tier Features

Voice transcription features vary by subscription tier:

Feature	Student	Essentials	Pro	Enterprise
Batch transcription	Yes	Yes	Yes	Yes
Streaming + live notes	Yes	Yes	Yes	Yes
Live clinical guidance	--	--	Yes	Yes
Session memory	--	--	Yes	Yes
Multi-speaker (couples/groups)	--	--	Yes	Yes
Growth analytics	--	--	Yes	Yes

Telehealth Integration

Video Sessions

Voice transcription works during telehealth sessions (Jitsi, Zoom, or Coviu) just as it does for in-person appointments:

1. Transcription captures both sides of the conversation
2. Audio is routed from the video platform automatically
3. The same consent requirements apply
4. Note generation works identically to in-person sessions

Hybrid Sessions

For mix of in-person and video:

1. Configure audio sources
2. Ensure all participants audible
3. Test before session

Settings and Customisation

Audio Settings

Settings > Clinical > AI Features > Voice

• Input device selection
• Sensitivity adjustment
• Noise cancellation toggle
• Audio quality level

Transcription Settings

• Auto-start preference
• Language selection (Australian English default)
• Medical vocabulary enhancement
• Timestamp granularity

Privacy Settings

• Retention period
• Auto-delete option
• Transcript storage location
• Consent workflow

Keyboard Shortcuts

Shortcut	Action
Ctrl+M	Toggle transcription
Ctrl+P	Pause/resume
Ctrl+K	Add marker
Ctrl+G	Generate note

Tips for Effective Use

1. Test before sessions - Run an audio test before each client arrives to ensure your microphone is working correctly
2. Explain to clients - Transparency builds trust; explain how transcription helps you provide better care
3. Review transcripts promptly - Catch errors while the session is fresh in your memory
4. Use markers liberally - Flag important moments during the session for easy reference later
5. Keep backup notes - Have pen and paper ready in case of technology issues
6. Position your microphone - Place it equidistant from you and your client for balanced audio capture
7. Minimise background noise - Close windows, turn off fans, and silence phone notifications
8. Speak naturally - Overly slow or exaggerated speech can reduce transcription accuracy

Related

• Note Assistance
• Creating Notes

Was this page helpful?