Extract text from speech, screenshots, and files. Supports real-time voice recognition, screen capture OCR, and batch processing of audio, video, images, and PDFs.
Getting Started
Choose one of three modes from the main screen:
- Live Transcription — real-time speech-to-text from your microphone
- Capture Transcription — OCR from screenshots or camera
- File Transcription — batch process audio, video, image, or PDF files
Live Transcription
Real-time speech recognition from your microphone or audio input device.
Speech Engines
| Engine | Type | Best for |
|---|---|---|
| Google Speech | Online API | High accuracy, many languages (default) |
| Whisper | Local model | Offline use, privacy (tiny to large models) |
| Vosk | Offline | Lightweight, fast, many languages |
Recording Options
- Audio input device selection
- Audio format: MP3 or WAV
- Audio quality: low / medium / high
- Optional video capture with camera preview
- Optional speaker classification (identifies different speakers)
Capture Transcription
Extract text from screenshots or camera captures using OCR.
- Camera capture — select a camera device, take a single shot
- Full screen capture — capture the entire screen
- Region selection — draw a rectangle to capture a specific area
- Window capture — capture a specific window
OCR engine: Tesseract (Japanese by default). Results appear alongside the captured image.
File Transcription
Batch process files by dragging and dropping or using the file browser.
Supported Formats
| Type | Formats | Method |
|---|---|---|
| Audio | MP3, WAV, M4A, FLAC, AAC, OGG, WMA | Speech engine |
| Video | MP4, AVI, MOV, MKV, WMV, FLV | Audio extraction → speech engine |
| Image | JPG, PNG, BMP, TIFF, GIF | Tesseract OCR |
| Document | Page → image → OCR |
Workset History
- Each transcription session is saved as a workset
- Filter by keyword, type (Live/Capture/File), and status
- Double-click to reopen a previous session
- Sortable columns: name, type, method, date, status
Settings
| Setting | Description | Default |
|---|---|---|
| Recognition Model | Default speech engine | |
| Audio Format | Recording format | WAV |
| Audio Quality | Compression level | Medium |
| Video Capture | Enable camera recording | Off |
| Speaker Classification | Identify different speakers | Off |
Tips
- Offline transcription — Use Whisper or Vosk for completely offline speech recognition
- Meeting notes — Enable speaker classification to distinguish who said what
- Batch OCR — Drop an entire folder of scanned documents for batch processing
- PDF extraction — Each PDF page is converted to an image and OCR'd individually
Privacy
- Whisper and Vosk engines run entirely on your device
- Google Speech sends audio to Google servers for recognition
- All transcription results and recordings are stored only on your local disk
- No analytics or tracking