User Guide

Quick Start

1. Open the app and complete the welcome setup if this is your first launch.
2. Choose Private first, Balanced, or Best accuracy.
3. If offered, tap Install Bundled Defaults to install the included Moonshine and Sherpa local models.
4. Grant microphone permission only if you want to record inside the app. Importing files does not need microphone access.
5. Start with Record for a live recording or Import for an existing audio file.

Record Audio

1. Tap Record in the bottom bar.
2. Choose the transcription style or engine you want to use.
3. Tap Start Recording.
4. Use Pause, Resume, and Stop as needed.
5. When the recording finishes, the app saves the audio and starts transcription in the background.

If you see a message that the selected engine is unavailable, open Settings or Advanced Engine Manager and check whether that model or API key still needs to be added.

Import Audio

1. Tap Import in the bottom bar.
2. Stay on the Setup tab and choose a transcription style.
3. Tap Choose Audio File.
4. Pick an audio file from Android's file picker.
5. Watch progress under the Live tab, or open Queue to inspect previous import jobs.

Imported files are transcribed in the background, so you can leave the screen and come back later from Library or Import Queue.

Use Local Models

The app supports four on-device engines: Moonshine, Sherpa-ONNX, Whisper local, and Vosk. Moonshine and Sherpa are the main built-in local options. Whisper and Vosk can be downloaded in-app when you want them.

Install or Download a Local Model

1. Tap the top-right Settings icon.
2. Review Transcription Settings, then tap Open Advanced Engine Manager.
3. Open the Local Models tab.
4. Tap Install Bundled Defaults to install the included Moonshine and Sherpa models.
5. For Vosk, tap Download Model.
6. For Whisper Local, tap Download Model or Import File if you already have the model file.
7. Use Remove Installed if you want to free up storage later.

Choose Which Local Engine Runs First

1. Go back to Transcription Settings.
2. Use Recommended mode if you want a simple preset.
3. Or choose a specific Default engine and tap Use first.
4. If needed, open Advanced Engine Manager and use Engine Rules to change fallback order.

Add OpenAI or Deepgram API Keys

Cloud engines are optional. You only need to set them up if you want OpenAI or Deepgram transcription.

1. Tap the top-right Settings icon.
2. Tap Open Advanced Engine Manager.
3. Open the Cloud Setup tab.
4. For OpenAI, enter your OpenAI API Key. Adjust OpenAI Base URL only if you intentionally use a compatible custom endpoint.
5. Tap Test OpenAI.
6. For Deepgram, enter Deepgram API Key, then review Deepgram Base URL and Deepgram Model.
7. Tap Test Deepgram.
8. Tap Save Settings before leaving the page.

Once a key is saved, the related cloud engines become selectable in Transcription Settings, Record, and Import.

Open, Search, and Edit Transcripts

1. Tap Library.
2. Use Search titles and transcripts to find text inside old recordings.
3. Use the Filter chips to show running, queued, failed, or completed jobs.
4. Tap any recording card to open its detail screen.
5. On the Transcript tab, edit the transcript directly and tap Save.
6. Use the player at the bottom to scrub, play, pause, jump by 10 seconds, or change playback speed.

Use Time Ranges

1. Open the Time Ranges tab inside a recording.
2. Tap a time range to jump playback to that section.
3. Choose an engine if you want to retry just one section.
4. Tap Refresh Time Range to retranscribe only that selected segment.
5. Tap Save Segment Edit if you manually corrected the text.

Retrieve and Export Transcripts

Every completed recording or imported job lands in Library. To retrieve a transcript later, open the item from the library and use its transcript or export actions.

1. Open the recording from Library.
2. Stay on the Transcript tab.
3. Tap More Actions.
4. Choose one of the export options.

After choosing an export type, Android will ask where you want to save the file. Pick a folder or cloud drive destination, then confirm the filename.

Troubleshooting