VocalFuse Documentation — Local Dictation & Pro Note Taking

Overview — local speech to text on Windows

VocalFuse is Fuse Intelligence’s local-first speech-to-text desktop app (VocalFuse.exe). A ~~57 MB transcription model runs on your PC — no cloud speech API, no meeting bot. Subscriptions, product keys, downloads, and saved notes live on this website. For marketing comparisons see VocalFuse features, AI note taker, or pricing.

• Basic ($5/mo): Unlimited local hold-to-speak dictation (Transcribe mode), always-on-top pill, product key, in-app updates — audio never uploaded
• Pro ($10/mo): Everything in Basic, plus Note Taking (click start/stop, AI summaries), Continuous (60-second background clips, local only), optional Account → Notes sync, optional Gmail summary delivery (notes_email_to)
• Model: A transcription model (~57 MB) is downloaded on first launch into C:\VocalFuse\models\ and stored locally
• Privacy: Microphone audio is transcribed locally. Note Taking uploads text only (transcripts + summary) to your account — not raw audio

Requirements

Grant microphone permission in Windows Privacy settings. Dictation pastes into the focused app via clipboard injection — keep the destination window focused (the pill never steals focus).

Install

1. Sign in on the website and open an active VocalFuse subscription on Subscriptions.
2. Copy your key from Product Keys.
3. Download VocalFuseInstaller.exe from Downloads.
4. Run the installer, paste the product key when prompted, and finish setup (app + updater land under your install folder).
5. On first launch, VocalFuse downloads the transcription model (~57 MB) into C:\VocalFuse\models\ if needed.
6. Confirm the floating pill appears. Open Settings (gear) if you need to re-enter the key later.

Product key & APIs

Your product key links the desktop app to your subscription. Paste it in the installer or in Settings → Product License. After the key and app version are accepted, the desktop loads that account’s profile: name, email, plan (Basic or Pro), and the “store notes on website” flag. Note Taking / Continuous stay hidden on Basic; they require a verified Pro key. Transcribe works without a key for local dictation only.

If verification fails, confirm the subscription is active and that HTTPS to https://fuseintelligence.org/api/license/verify.php is allowed.

After a session ends, a second upload uses kind: "summary" with the AI subject/body. Open everything in Account → Notes.

Recording modes

Choose a mode from the pill’s mode dropdown or Settings → Recording Mode. There is no global hotkey chord — you interact with the pill itself.

Mode picker

Pill overlay UI

The floating overlay stays on top without activating itself (WS_EX_NOACTIVATE), so your target app keeps keyboard focus for paste. Drag the left grip to reposition — placement is remembered per monitor.

Pill states

Settings panel

Opened from the pill gear. Collapsible cards cover license, recording mode, email delivery, and updates. There is no Performance / CUDA settings card in the current build.

Note Taking mode (Pro)

Note Taking produces summarized notes from long transcriptions — meetings, calls, or videos. It is the Pro workflow that fills Account → Notes. Audio never leaves your machine for transcription; only text is uploaded.

1. Confirm Pro on Subscriptions and paste a verified product key in Settings.
2. Choose Note Taking from the pill mode menu (label shows “Notes”).
3. Click the pill — status becomes “Taking notes…”. Speak through the meeting, call, or video.
4. Speech is transcribed locally and saved into C:\VocalFuse\Notes\<session_id>.json. If cloud notes are on, each transcript uploads for that session.
5. Click the pill again to stop. The app builds an AI summary of the long transcription, writes it into the same local JSON file, uploads it as a summary note (if cloud storage is enabled), then optionally emails it.
6. Refresh Account → Notes to open the session panel (summary + transcripts together).

• Session ID groups every batch + the final summary into one web panel.
• If the AI summary call fails, VocalFuse falls back to a local bullet list from the transcript text — still uploaded.
• Email delivery and cloud storage are optional Pro preferences on Account Settings. The desktop app reads them when it verifies your product key. Local JSON notes still save under C:\VocalFuse\Notes\.

Account → Notes (website)

Signed-in users open Account → Notes to review everything Note Taking uploaded for their license key.

• Each session is one expandable dropdown panel (collapsed by default).
• The panel title comes from the AI summary subject when present.
• Inside: AI summary first, then numbered transcript segments for that session.
• Timestamps display like Saturday September 2nd 2023.
• Edit the summary (or a segment) from the panel; Delete removes the whole session.
• Your linked product key is shown at the top of the Notes page when available.

Continuous mode (Pro)

Continuous records in the background and checks 60-second clips for anything important: click to start, click to stop. Matching clips become local bullets under %AppData%\VocalFuse\continuous_bullets.json. When you stop, VocalFuse can email a final summary — those notes stay on your PC and do not sync to Account → Notes.

Use Continuous when you want the app listening in the background for important moments; use Note Taking when you want summarized notes from a long transcription (meeting, call, or video).

Updates

Settings → Updates compares your installed build with the Fuse API version and can launch the updater. Installers and VocalFuse.exe also ship from Downloads. User-facing changes are listed on the changelog.

This documentation targets release 1.0.18. Restart VocalFuse after updating.

Troubleshooting

No text appearing at the cursor

Keep the destination app focused before you release the pill. The overlay never steals focus. Confirm the app accepts pasted text.

Notes missing on the website

Use Note Taking (not Continuous), with a verified Pro key and network access to fuseintelligence.org. After stopping, wait for “Sending notes…” to finish, then refresh Account → Notes.

Summary looks like raw bullets, not a real summary

That is the offline fallback when the AI summary request fails. Check internet access and retry a short Note Taking session on the latest app version.

“Invalid key” or license errors

Renew on Subscriptions, re-copy from Product Keys, and allow HTTPS outbound.

Note Taking / Continuous greyed out

Those modes need a verified Pro product key. Transcribe still works without one.

Model missing or failed download

Confirm the model finished downloading into C:\VocalFuse\models\ (about ~57 MB), then restart VocalFuse.exe.

Microphone not detected

Check Windows Privacy → Microphone, the default input device, and that no other app holds exclusive mic access.

Email summary never arrives

Set Email summaries to on Account Settings, then restart or re-verify the product key in VocalFuse. Leave it blank if you only want Account → Notes. Cloud sync is a separate toggle on that same page.

Related