Fikr Vox user guide

Install Fikr Vox, set up a voice, and make a voiceover, podcast, or dubbed video.

Install Fikr Vox

Fikr Vox runs on Apple silicon Macs with macOS 13 or later. Check GitHub Releases for packaged builds.

Run from source

Install Node.js 22 or newer, Rust, Xcode with the macOS 26 SDK or newer, Python 3.11 or newer, and uv on your PATH. You'll need internet access to install dependencies and download voice models. Tauri prerequisites · Install uv

git clone https://github.com/bigmints/fikr-vox.git
cd fikr-vox
npm ci
cargo fetch --locked --manifest-path src-tauri/Cargo.toml --target aarch64-apple-darwin
npm run desktop

Build a local app bundle

npm run build:desktop

The command creates src-tauri/target/release/bundle/macos/Fikr Vox.app. Open the bundle from that directory or copy it to Applications.

Create your first project

  1. Open Settings → Connections → Audio and download Kokoro for preset voices on macOS 14 or later. On macOS 13, connect a cloud speech provider.
  2. Choose New and select Story Telling, Podcast, or Video Dubbing.
  3. Write or import a script, select a voice, and generate a take. Listen to the result before exporting.

Voices and connections

Local voices

Downloaded voice models require macOS 14 or later. Open Settings → Connections → Audio to download a model. Kokoro provides preset voices. Qwen3 offers voice cloning, presets with delivery controls, and voice design from a description. The first download installs a pinned Python 3.11 runtime. Generation runs offline after setup.

To clone a voice, open My Models, create or edit a model, and choose a Qwen3 cloning engine. Import or record a clean voice sample and enter its transcript. Use recordings and portraits you have permission to use.

Transcription and translation

Download Whisper under Settings → Connections → Transcription to turn recordings into text on your Mac. Open Translation to set up Compact multilingual or Apple Foundation Models. Apple's option requires macOS 26 and an eligible Mac with Apple Intelligence enabled.

Cloud providers

Open Settings → Connections and select Connect beside OpenRouter. Enter your key, then select Refresh models in the relevant connection group. Choose a voice or transcription model when editing a profile under My Models. Capabilities vary by model. Your provider may charge for requests.

Produce a podcast

  1. Open Speakers & voices and assign a voice to the host and each guest. A project supports up to seven guests.
  2. Choose Import script and use labels that match the speaker names.
Host: Welcome to our first episode.
Guest: Today we're talking about how this project got started.
  1. Select Generate voices. Matching completed takes can be reused. Stop after current turn lets the active take finish and stops further submissions.
  2. Choose a conversation timing preset and review the dialogue. Adjust track volume and mute settings, then select Export audio for a mixed WAV file.

Dub a video

  1. Create a Video Dubbing project and import an MP4, MOV, or WebM. H.264 in MP4 or MOV is a good choice for macOS.
  2. Choose a voice profile and write the narration. You can also transcribe the source video and edit the transcript. Select Generate twin voice when the script is ready.
  3. Set the start time and choose Keep natural pace or Fit to remaining video. Fit can adjust voice speed from 0.5× to 2×.
  4. Select Render dubbed preview. Review the result and choose Download dubbed MP4.

Dubbing replaces the original audio and preserves the video's length, picture, and lip movements. Source videos can be up to 512 MB and two hours long. Source transcription supports up to ten minutes and 20 MB of extracted WAV audio.

Storage and privacy

New desktop installations store projects, media, and voice models in ~/Library/Application Support/com.bigmints.fikr.vox/. Existing installations can continue using com.pretheesh.mytwin when that folder contains the project database. The browser development server uses a separate data/ directory in the checkout.

Provider keys are stored locally in an unencrypted file with owner-only permissions. Cloud actions send the required inputs to the selected provider. Remove credentials and personal media before sharing logs or project backups.

Troubleshooting

A local model isn't ready

Check the download under Settings → Connections → Audio. If setup fails, check your connection and available disk space.

A cloud request fails

Check the provider key, credit balance, and model capabilities. Review provider activity before retrying an interrupted paid request.

Narration doesn't fit the video

Move the start time earlier, shorten the script, or choose Fit to remaining video. For import or rendering errors, try an H.264 MP4 and check that the Xcode command-line tools are installed.

Include your macOS version, app version or commit, and steps to reproduce the problem when asking for help. Open a GitHub issue

Contribute

The desktop app uses Rust and Tauri. The desktop and browser versions share an HTML, CSS, and JavaScript interface. Development and testing guide

License

Fikr Vox's source code is available under the MIT License. Third-party libraries, models, and connected services have their own terms. Read the MIT License