Short, honest answers about how VoiceJar works, what runs offline, which AI providers it talks to, and where your recordings live. Looking for pricing specifics? See the Pricing page. Want the data-and-keys deep dive? See the Privacy page.
Start here if you’ve never used the app.
VoiceJar is a voice-first note-taking app. You tap once to record — from a button in the app, from the notification, or from the Android floating bubble — and the app handles the rest: transcribes the audio, writes a summary and title, suggests tags, and extracts any tasks it heard. Everything is filed into an Inbox where you can organize later or never at all.
Three concrete differences. One — VoiceJar is BYOK-friendly: you can bring your own keys for OpenAI, Anthropic, Google Gemini, Groq and ElevenLabs and pay the provider directly. Two — the capture-and-transcribe core runs fully offline with Whisper.cpp on-device; no internet needed for the basic loop. Three — no signup, no payment, no email is required to start — you can install the app and be recording in under a minute.
No. First launch auto-starts a 14-day Pro trial without any signup, payment or email step. You can use VoiceJar indefinitely without ever creating an account. If you later subscribe to Pro on the website, Stripe emails you a license key that works on Android and Windows.
The recording → transcription core is fully offline once you download a Whisper model in Settings (tiny … large-v2 are available). The app even acquires a CPU wake-lock so long Whisper jobs don’t get throttled by Android Doze. AI steps that need a cloud LLM (summary, title, tags, task extraction) are queued and run when you reconnect.
VoiceJar talks to several engines. You pick which one and you supply the keys.
Transcription (4): Whisper.cpp offline, plus three cloud BYOK options — OpenAI Whisper, ElevenLabs Scribe, and Groq Whisper. Text / summarization (4 LLM providers): Anthropic (Claude), OpenAI (GPT), Google Gemini, and Groq. You can mix providers across steps — for example, Groq for fast titles, Claude for the deep summary, Gemini for tag suggestions. ElevenLabs and OpenAI also power optional read-aloud (TTS).
If a cloud transcription call fails and you have a Whisper model downloaded, VoiceJar automatically retries the request on-device and flags the note so you know it ran offline. Failed cloud calls do not burn your daily cloud quota — only successful calls count.
No — this is the one feature that always needs the internet. The translator captures speech, translates it with a cloud LLM, then speaks the result aloud through a cloud text-to-speech voice (ElevenLabs or OpenAI). Both stages are cloud calls, so live translation does not work on a plane. Without a TTS key you still get the translated text on screen, just not the spoken playback. It draws on the same daily cloud allowance (5/day free, uncapped on Pro), and finished conversations are saved on-device so you can reopen the transcript later.
“If we can’t verify it against the source code, it doesn’t ship on this page.”
Two surfaces that only exist on Android, and one legal disclaimer.
On Android, yes — via integration with the third-party app BCR (Basic Call Recorder). VoiceJar does not record calls itself; it imports the audio file BCR produces, attaches a header with the contact name, phone number, direction and duration, and runs the normal transcription & AI pipeline. You’ll need to install BCR separately. iOS does not allow third-party apps to capture call audio. Local laws on call recording are your responsibility.
On Android, VoiceJar can show a small system-wide bubble that floats
on top of any other app. Tap it to start or stop a recording without
switching to VoiceJar. It uses the Android SYSTEM_ALERT_WINDOW
permission (you grant it once in Settings). The bubble is Android-only
— iOS does not allow system overlays.
Projects, tags, tasks, 14 UI languages, merge, export. No lock-in.
Every recording lands in Inbox (no project assigned)
and you can organize later, never, or whenever you want. Projects are
flat (no nested sub-projects), each with a custom color and emoji.
Tags are AI-suggested from your existing taxonomy, so VoiceJar
won’t spam you with new tags every time. Tasks are AI-extracted
from transcripts: it parses {title, dueDate} from the text
(“tomorrow”, “next Monday” resolve correctly) and you can opt
into a daily reminder for pending tasks.
The app interface is fully localized into 14 languages (English, Slovak, Czech, German, French, Spanish, Portuguese, Italian, Hungarian, Polish, Japanese, Chinese, Russian, Ukrainian). Transcription accepts any language Whisper or your chosen cloud provider supports — not English-only. AI prompts respect the language you set for the note.
Yes. Multi-select in the notes list, choose “Merge”: the audio files are joined with a 1-second gap, transcripts are concatenated, and you get a single note. Merge is irreversible (the app warns you) and is available on every tier, including free offline.
Yes — per-project export is in the offline feature set, meaning it works on every tier. You can pull notes, transcripts and audio out of VoiceJar at any time. There is no lock-in.
The three things people ask before installing.
Locally on your device. Audio files live in app-private storage
(Android / iOS sandbox), and notes / projects / tags / tasks live in
an on-device SQLite database (Drift). Your API keys live in the OS
Keychain (iOS) or Keystore (Android) via flutter_secure_storage.
Nothing is uploaded to VoiceJar servers as part of normal use. See the
Privacy page for the deep dive.
Every install starts with a 14-day Pro trial, optionally extendable by 30 days when you opt in to product tips with your email. After that you can stay on BYOK Free Forever (use your own API keys, plus a daily allowance of 5 cloud transcriptions on us), or upgrade to Pro (from $4.99 / month through Stripe) which removes the 5/day cloud cap, adds priority support, and provides one license key for up to three Android or Windows devices. The app never blocks on payment failure — it soft-lands to BYOK Free. Full breakdown on the Pricing page.
Android is live on Google Play (requires Android 7.0 / Nougat or newer), and Windows 10/11 is available as a direct x64 download. iOS is coming soon; features like the floating bubble and call recording aren’t possible there. See the Download page.
Didn’t find your answer? Drop us a line — we read everything.
14 days of Pro on first launch — no signup, no payment, no email. At the end, optionally opt in to product tips with your email for +30 more days — still free. Decide later if Pro is worth a few dollars a month.
Google Play · Free 14-day Pro trial
Download for Android →Android 7.0+ · iOS coming soon