Guide
Recording and dictation
With the client's consent, Neurova Practice records the session, makes the transcript on your Mac and drafts the note under your headings. You read it, correct it and sign it. The audio is deleted as soon as the transcript is made.
On this page
- What you need
- Spoken language
- Recording consent
- Record a session in person
- After the session: transcript and draft
- The AI draft is marked
- The transcript panel
- Dictate after a session
- Quick dictation from the menu bar
- Dictate on iPhone and iPad
- If the app quits during a recording
- When the practice turns recording off
What you need
Two models run on your Mac. Set them up under Settings → General, under Voice notes:
| Setting | What it does | Size |
|---|---|---|
| Speech recognition | Turns speech into text (Parakeet v3) | About 0.5 GB |
| Draft model | Writes the draft note. Standard (Gemma 4 E4B) or Compact (Gemma 4 E2B) | About 5.2 GB or 3.6 GB |
Click Download next to each. The row shows Ready when it’s done. Use Compact on Macs with 16 GB memory or less.
The speech model is fetched by itself the first time a recording needs it. The draft model is never downloaded without you: until it’s there, you get a transcript but no draft.
Both run entirely on your Mac. Audio and transcripts never leave it.
Spoken language
Spoken language is set to Automatic by default. It recognises 25 European languages, whatever language the app itself is in. Choose one language only if a recording comes out in the wrong alphabet.
Recording consent
Sessions are recorded only for clients who gave consent. Record it in the client file:
- Ask the client for consent before the first recorded session. Explain that the audio is deleted right after the transcript is made, and that they can withdraw at any time.
- Open the client file. On the Overview tab, under Recording, click Record consent given today.
The file then shows Given on with the date. Click Withdraw if the client changes their mind. From then on, their sessions aren’t recorded.
Record a session in person
- Open the appointment and click Start session. Under the button it says whether the session will be recorded.
- The session window opens. With consent, recording starts straight away and a red REC badge shows next to the timer.
- Talk with your client. Use Pause recording and Resume recording for parts that shouldn’t be in the transcript.
- Click End session and confirm.
The session window is kept quiet on purpose. It shows the client’s name, a timer, the recording badge, what was in the last session’s signed note, and a Scratch pad. There’s no live transcript while the client is with you.
Use the scratch pad for your own notes during the session. Only you see it. It’s used for the draft, and erased when the note is signed.
Without consent, the badge reads Not recording: no consent, and you can still take notes.
Not recording this session
A client who gave consent may still not want a particular session recorded. With consent, and when your practice allows recording, the session window has a Record switch next to the timer. Switch it off: the recording stops, and what was recorded in this session so far is deleted. The access log notes that the recording was deleted. In a video session, the call’s recorded tracks are deleted the same way.
Video sessions are recorded from the call itself, one track per person. Recording a video session.
After the session: transcript and draft
When you end the session, the appointment is marked Done and the note opens. Above the headings you see the progress:
- Making the transcript on this Mac… The recording is turned into text. The audio file is deleted right after.
- Drafting the note on this Mac… The draft model writes under each heading of the note’s template, if the draft model is Ready.
You can keep working meanwhile. If you already typed under a heading, the draft leaves it alone: it only fills headings that are still empty.
The draft model is told to use only what was said and what’s on your scratch pad. It doesn’t give a diagnosis, a risk assessment or treatment advice, and it only records agreements made in the session. A heading with nothing to go on reads Not recorded in the transcript.
The AI draft is marked
While the text is still the draft as the model wrote it, the note says so:
This text is an AI draft made from the transcript. Check every heading against what was said, and correct it, before you sign.
If you try to sign the unchanged draft, the confirmation warns you again. The note’s history records which version the model made. Once you change the text yourself, it’s your version.
You always read, correct and sign the note. The AI only drafts documentation.
The transcript panel
Click Transcript in the note’s toolbar to show or hide the panel beside the note. It shows the transcript, and has:
- Dictate and Stop dictation, to add to the transcript by speaking;
- Draft from transcript, to draft the note from the transcript by hand. It needs the draft model to be Ready, and like any draft it only fills headings that are still empty. Until then, the panel says Download the draft model in Settings → General → Voice notes first.
Transcript and scratch pad are erased when the note is signed.
Dictate after a session
No recording, or something to add? Dictate your own summary instead:
- Open the note and click Dictate in the transcript panel.
- Speak your summary. The panel shows Recording your dictation….
- Click Stop dictation.
The dictation is turned into text on your Mac and added to the transcript. The audio is deleted. If the note’s headings are still empty and the draft model is ready, a draft follows.
Quick dictation from the menu bar
Neurova Practice sits in the menu bar too. Press ⌃⌥D anywhere on your Mac, or choose Quick dictation (⌃⌥D) in the menu.
- Check the client. The client of the session under way, or starting within 15 minutes, is chosen for you.
- Click Record, or press ⌃⌥D again.
- Speak, then click Stop and save, or press ⌃⌥D.
The memo becomes a new note for that client, with your default template. It’s turned into text on your Mac and the audio deleted right after, then drafted like any recording. ⌃ is Control and ⌥ is Option.
Under Settings → General, under Menu bar, you choose whether the app shows in the menu bar, whether the menu shows client names or initials only, and whether it opens at login. Client names never appear in the menu bar itself.
Dictate on iPhone and iPad
Open a note that isn’t signed and tap Dictate. Tap the microphone, speak your summary, and tap stop. The speech is turned into text on the device: the first time, it gets the speech model ready (about 0.5 GB, once). The audio is deleted right after, and the text is added to the note.
Your Mac turns it into a draft under the headings when you open the note there.
If the app quits during a recording
A recording left behind by a crash or a forced quit isn’t lost. The next time Neurova Practice opens, it’s transcribed into its note and then deleted. If that note was signed in the meantime, the recording is deleted without being used.
When the practice turns recording off
A practice on its own can switch recording off under Settings → General: under Practice, switch off Allow recording sessions and dictation.
In a team, the owner decides whether anyone may record. Under Settings → Team, in Device rules, switch Allow recording sessions and dictation off, and click Save rules.
Sessions are then not recorded on any of the practice’s devices, and dictation, quick dictation and dictation on iPhone and iPad stop with the message Your practice has turned off recording and dictation.
Last updated 19 November 2026
Can't find it here? See Support and FAQ or email support@neurova-ai.com. We reply by email, usually within two working days.