Voice Transcriber

Review-first Linux dictation · x86_64 Flatpak · Groq

Dictate the draft.
Keep the final say.

Give spoken prompts, emails, tickets, and notes a review step before they reach another app. Voice Transcriber filters silence locally, makes the cloud boundary visible, and leaves the transcript editable.

Download for Linux Watch the workflow

Review before copy · no app-created audio files · no analytics · history off by default

Guided preview · synthetic sample text

VOICE TRANSCRIBER

LIVE SESSION

■   Stop listening

Listening · speech detected (3.2s)

INPUT SIGNALUSB microphone

“The important part is getting the idea out before it disappears.”

Keep on top  ●Clear   Copy

The complete loop, before you install

Speak. See the state. Shape the result.

This short, reproducible tour mirrors the current native GTK states. The app itself never inserts a simulated transcript.

Guided Voice Transcriber workflow moving from ready, to active listening, to an editable completed transcript
Ready → listening → review. Synthetic sample text keeps the public demo reproducible and privacy-safe.Reproduce the tour ↗
Voice Transcriber ready state with microphone and privacy information visible
01 · Inspect. See the selected device and provider boundary before configuring a key.
Voice Transcriber active state with speech detected and input meter moving
02 · Speak. Local voice detection and the input meter make capture state visible.
Voice Transcriber completed transcript ready for review, copy, or export
03 · Review. Edit the result, then copy, clear, or explicitly export it.

A deliberate checkpoint

Built for the moment most dictation tools skip.

Voice Transcriber does not type invisibly into the active window. It gives you a focused place to catch the thought, correct the words, and decide where the final text belongs.

Reach for it when

  • You are drafting a prompt, email, ticket, note, or message.
  • You want to review wording before it reaches another app.
  • You prefer cloud speed without downloading a multi-gigabyte model.
  • You want the provider boundary visible instead of buried in policy copy.

Choose another tool when

  • Audio must remain fully offline.
  • You need a global hotkey while another app has focus.
  • You want automatic cursor insertion or meeting transcription.
  • You require Flathub discovery, ARM, macOS, or Windows today.

One bounded path

From microphone to useful text.

  1. 01

    Inspect

    Choose a microphone and confirm its local signal before any session starts.

  2. 02

    Speak

    Local voice detection filters silence and closes each completed speech segment.

  3. 03

    Review

    Pending, completed, and failed segments remain visible while the text stays editable.

  4. 04

    Move it

    Copy the finished desk or export plain text, Markdown, or a timestamped note.

Privacy-explicit, not “offline”

Fast cloud transcription. An exact local boundary.

Stays on your device

  • Raw microphone frames in a bounded memory queue
  • Silence detection and the live input meter
  • The transcript until you copy or explicitly export it
  • History unless you opt in to local text storage

Sent when you speak

  • Each completed speech segment, encoded in memory as WAV
  • The selected language or translation request
  • Nothing while the app is waiting in silence
  • No app analytics, crash reports, or transcript telemetry

Read the provider facts, storage rules, and complete data flow

Proof and limits, side by side

Trust the evidence—not a badge.

69deterministic tests pass without a key, microphone, model, or network
v1Flatpak with checksum, test report, CycloneDX SBOM, and Sigstore provenance
0analytics or crash-reporting pipelines inside the application

Still open:

  • Physical PipeWire/PulseAudio plus Wayland/X11 compatibility reports
  • Flathub distribution and automatic application updates
  • A supported packaged local provider and portal-backed global shortcuts

Inspect the v1 release ↗Read the support matrix ↗Report a Linux setup ↗

Checksum-verifiable x86_64 Flatpak

Download it. Verify it. Inspect before adding a key.

Download v1.0 Flatpak View all release evidence ↗

terminal

flatpak remote-add --user --if-not-exists flathub https://flathub.org/repo/flathub.flatpakrepo
curl -fLO https://github.com/OthmaneBlial/audio-capture/releases/download/v1.0.0/voice-transcriber-1.0.0-x86_64.flatpak
curl -fLO https://github.com/OthmaneBlial/audio-capture/releases/download/v1.0.0/voice-transcriber-1.0.0-x86_64.flatpak.sha256
sha256sum --check voice-transcriber-1.0.0-x86_64.flatpak.sha256
flatpak install --user ./voice-transcriber-1.0.0-x86_64.flatpak
flatpak run io.github.othmaneblial.audio_capture

The interface, device picker, diagnostics, and privacy explanation work without a key; Start listening stays blocked until Groq is configured. This standalone bundle does not create an update remote. Read the full install/update guide or use the source development path.

Useful on your Linux setup?

Star it—and tell us where it ran.

Stars help the right people discover the project. A privacy-safe compatibility report helps make the next release genuinely better.

Star on GitHub Report your setup