Using AI: live audio mode (Beta)

Similia's Live Audio Mode (Beta) lets you record a consultation in real time while the AI:

• Transcribes the conversation as it happens
• Picks out the symptoms the patient reports: chief complaint, sensations, modalities, causations, mental and emotional qualities, and Strange, Rare & Peculiar (SRP) symptoms
• Adds the best-matching rubric for each symptom to your analysis, from the premium repertory you select
• Surfaces Materia Medica findings and a running case analysis with suggested next questions

Everything after the transcript runs only in Transcribe + AI analysis mode (see "Session modes").

Availability: Pro-only and uses AI credits. Automatic rubrics also need a premium repertory on your account (Murphy, Suggesta, Complete, Saine, Meditative or Griffith). Pro Base and Lifetime Pro do not include one on their own; without one, Live Audio still transcribes and detects symptoms but adds no rubrics. Live Audio also needs your AI-processing consent, microphone permission, an internet connection (Live Audio does not run while you work offline), and the explicit permission of every person being recorded. If you have not given AI-processing consent yet, Similia asks for it after the microphone prompt (and, for Online, the sharing dialog). Once you accept, recording starts. In Online mode the sharing dialog opens again, so share the call once more.

How to use it (short version)

1. Open a case and create or open an analysis.
2. Click Live Audio (microphone icon) in the analysis toolbar.
3. Each time you open Live Audio, Live Audio Mode - How It Works and then Settings open by themselves, until you tick Don't show again (How It Works) or Don't show this again (Settings). Closing them (Got it, Confirm) does not start a recording.
4. In Settings (gear icon), choose the Language spoken in the consultation and a Session mode (see below). For Transcribe + AI analysis, also choose your sources: a Sources preset on the Simple tab, or a Repertory and the Materia Medica sources on the Advanced tab. Sources your account can't use can't be selected. Click Confirm.
5. Make sure everyone being recorded has given explicit permission.
6. Click the microphone button (Start Recording) in the session header.
7. In Before we start, choose:

   
   

   
   If this analysis already has a saved Live Audio session, Similia first asks Continue session (add to it) or Start fresh (discard it); see "If the page reloads or the screen goes blank". An AI Credits window may follow: click Continue to go on.
8. • In person (microphone only): records your microphone only (ideal for face-to-face consultations). Recording starts once you have answered any window that appears first (see below).
   • Online (microphone and call audio): records your microphone plus the call audio, so the transcript has both speakers. Online consultation setup opens first (see "Online consultations" below).
8. Allow the browser's microphone prompt if it appears. The header shows Recording... with a running timer, and transcript text appears as people speak.
9. Click the microphone button again (Stop Recording) to end the session. Similia writes the summary and saves the session to the analysis.

Online consultations

Choosing Online (microphone and call audio) opens Online consultation setup, which asks where the call is running. Pick the option that matches your call and click Continue. Allow the microphone prompt if it appears; the browser's sharing dialog opens next. The setup shows a picture of the Chrome dialog for each option, numbered to match these steps.

In a browser tab (Google Meet, Zoom web, Teams web):

1. Select the browser tab where the call is running.
2. Turn on "Share with tab audio" (older Chrome shows a checkbox called "Share tab audio").
3. Click Share, then keep the call tab unmuted.

In a desktop app (Zoom, Microsoft Teams, Skype, FaceTime):

1. Select Entire screen, not the app window: a single window carries no audio.
2. Turn on "Share with system audio" (older Chrome shows a checkbox called "Also share system audio"). This is what carries the patient's voice.
3. Click Share and keep the call app unmuted. Headphones are fine: the audio is captured before it reaches them.

Similia records audio only. The shared screen image is never stored or sent.

Browser requirements:

• Capturing call audio needs Chrome or Edge on a computer. In Firefox or Safari, and on a phone or tablet, Online mode records the microphone only, and the setup says so: "Call audio capture needs Chrome or Edge. In this browser only the microphone will be recorded."
• The desktop-app option needs system audio: Chrome or Edge on Windows, or Chrome 141 or newer on macOS 14.2 or newer. Where the browser cannot share system audio (for example an older Chrome on a Mac, or Linux), the setup shows a note under In a desktop app: join the call from a browser tab instead, or record the microphone only.
• On a Mac, macOS may ask once to let the browser record the screen and system audio (System Settings → Privacy & Security). Allow it, then share again.

If the call audio is missing:

• If the share comes back without audio, recording continues with the microphone only and a warning names what was missed, for example "A single window carries no audio. Share the entire screen with system audio, or the browser tab with “Share tab audio”. Recording microphone only." To fix it, stop the session and start again, sharing correctly.
• If you cancel the sharing dialog, or stop sharing during the call, the session continues with the microphone only.

Ticking Don’t show again in the setup skips it next time in that browser: choosing Online then opens the sharing dialog directly. There is no setting to show the setup again; the steps above are the same.

Session modes

In Live Audio's Settings, Session mode decides how much AI runs while you record:

• Transcribe only: a live transcript and a summary when the session ends. No symptom detection, rubric search, materia medica or case analysis runs during the session, so a session costs little more than the transcription minutes. The rubrics, materia medica and analysis panes step aside, leaving the transcript, with your notes beside it when Notes is open.
• Transcribe + AI analysis: the live transcript plus symptom detection, automatic rubrics, materia medica findings and a running case analysis.

The mode is locked while a recording runs ("Stop the recording to change the session mode."). The language and the repertory can't be changed during a recording either.

Manual work mode: if the app's Settings → Work mode is set to Manual, Session mode offers only Transcribe only, and every session runs that way. Choose AI-assisted under Settings → Work mode to bring back Transcribe + AI analysis; the session mode you had chosen on that device is restored.

Reading the transcript during the session

A Transcript toggle in the session header (Show transcript / Hide transcript) opens a resizable column with the full live transcript. It groups the text by speaker, follows the newest line as it arrives, and has a Copy button. On a phone the transcript is a tab.

Taking notes during the session

You don't have to close the session to write an observation. A Notes toggle in the session header (Show notes / Hide notes) — or the section toolbar's Notes button — opens the case notes beside the rubrics, materia medica findings and case analysis, with the same editor, formatting bar, templates and autosave as the Notes panel. Everything written during the consultation is already in the analysis when the recording stops. On a phone the notes are a tab. Each column stays open or closed the way you left it.

How rubric extraction works

In Transcribe + AI analysis mode:

• About every 2 minutes, the AI reads the latest part of the conversation and picks the observations the patient reports about themselves: the chief complaint with its location, sensations, modalities, causations ("never well since"), mental and emotional qualities, and Strange, Rare & Peculiar symptoms. Related details are combined into one symptom.
• The practitioner's questions and examples are ignored, and so is anything a rubric already in your analysis covers.
• For each symptom, the AI searches your selected repertory and adds the best match. It stays faithful to the patient's words, prefers a parent rubric over a narrower sub-rubric unless the patient described that detail, and avoids rubrics with so few remedies that they would force the result.
• Rubrics come from the premium repertory selected in Live Audio's Settings: Murphy, Suggesta, Complete, Saine, Meditative or Griffith. If none of these is available to your account, symptoms are still detected but no rubric is added.

Switching rubrics (when the AI is close, but not perfect)

When a rubric is added, you can click the ⇄ switch icon next to it to replace it with one of the top alternative rubrics suggested by the AI.

Session length and silence

A live session can run for up to 3 hours. Recording also stops automatically after 5 minutes without detected speech ("Recording stopped due to 5 minutes of silence."). In both cases the session is saved as if you had clicked Stop Recording.

Auto-reconnect and busy transcription

If the connection to the transcription service drops, Live Audio reconnects by itself: the header shows "Reconnecting (1/5)…", counting each attempt, and a "Connection restored" notice appears once transcription resumes. If it cannot reconnect, recording stops and the transcript is saved ("Connection lost. Maximum reconnection attempts reached. Your transcript has been saved.").

If the live transcription service is temporarily unavailable, Similia waits before retrying. Brief interruptions recover on their own; if the service needs a longer wait, recording stops and the captured transcript is saved ("Live transcription is temporarily unavailable. The session has stopped and its transcript is being saved."). Review what was captured before starting another session.

If the AI analysis cannot reach its service, the header shows "AI analysis paused — check your connection". The transcript keeps recording, and the analysis resumes when the connection recovers.

If the page reloads or the screen goes blank

While it records, Live Audio saves the session to the analysis every 10 seconds, so at most the last few seconds are lost. Open the same case and analysis and click Live Audio again. Similia then offers what fits:

• Incomplete Session Found: the session was saved without its summary. Generate Summary writes it now (uses AI credits); Later leaves it for another time.
• Previous live audio session (when you start recording): Continue session keeps the previous transcript, summary, materia medica findings and case analysis and adds to them in the new recording; Start fresh discards them and starts a new recording.
• Pending Summary (when you start recording while a summary is still pending): Generate & Start writes the summary first, then starts recording; Discard & Start discards the old session and starts a new recording.

Nothing was recorded?

• Closing Live Audio Mode - How It Works (Got it) or the Settings window that opens by itself (Confirm) does not start a recording. Click the microphone button (Start Recording) and choose an option in Before we start.
• If an AI Credits window appears when you start, click Continue. Closing it cancels the start, so nothing is recorded. Tick Don't show this again to skip it next time.
• While recording, the header shows Recording... and a running timer. Otherwise it shows Ready to Record.
• If the browser blocked the microphone, Similia shows "Microphone access denied. Please enable microphone permissions." Allow the microphone for Similia in your browser's site settings, then start again.
• With no AI credits left, Similia opens the credits window instead of starting.
• The audio itself is never saved, so a consultation that was not recorded cannot be recovered. Only what reached the transcript is kept.

Notes, credits, and privacy

• Always review suggestions: this feature is in beta; you make the clinical decisions.
• Credits: credits are consumed for transcription minutes and for the AI processing: the end-of-session summary in both modes, plus symptom detection, rubric picks and the case analysis in Transcribe + AI analysis mode. The credit counter in the session header follows the rate the session is billed at, and Settings shows a typical cost per hour before you start.
• Privacy: the audio recording is not saved. The analysis keeps text only: the transcript and the AI summary, plus, in Transcribe + AI analysis mode, the materia medica findings, the case analysis and the rubrics that were added. The saved session appears in the analysis notes under Live Audio Session. Turning AI processing off under Settings → Privacy & consent — even from another tab or device — stops a running recording and skips the final AI summary and analysis; the captured transcript stays available.

Related guides

• Repertorization and Analysis (Selecting Rubrics)
• Using AI: Case Notes to Rubrics
• Using AI: Photos to Rubrics
• Is my patient data secure?

Updated on: 11/10/2026