home
CapsGAT is a specialized transcription workstation designed for converting subtitle files into publishable interview transcripts using GAT2 (Gesprächsanalytisches Transkriptionssystem 2) conventions.
- Quick Speaker Assignment: Rapidly assign speakers to subtitle segments using keyboard shortcuts
- Basic Editing: Split, merge, and edit transcript segments
- GAT2 Symbols: Insert conversation analysis symbols for pauses, breathing, and annotations
- Audio Synchronization: Link audio files to transcripts with auto-sync functionality
- Export Transcripts: Generate formatted transcripts in HTML or plain text
- Start a new transcription project by selecting File → New Project or pressing Ctrl+N.
- Open saved projects with File → Open Project or Ctrl+O. CapsGAT uses the .capsgat file format.
- Save your work with File → Save Project (Ctrl+S) or Save Project As to create a new file.
Add project notes and descriptions via Edit → Project Memo. This information can be included in transcript exports.
- SRT (.srt): Standard subtitle format with timestamps
- JSON (.json): Various JSON formats including token-based and segment-based
- Text (.txt): Plain text files (one block per line)
- TSV (.tsv): Tab-separated values with start/end times and text
Use File → Import → Subtitles to load subtitle files. When importing audio files, CapsGAT automatically searches for matching subtitle files in the same directory and offers to import them.
When importing JSON files with token data, you can choose from three import methods:
- One Continuous Block: Import all text as a single segment
- Tokens as Separate Blocks: Create individual blocks for each token
- Auto-segment: Automatically detect pauses and create segments accordingly
Assign speakers using the number keys 1-4 (configurable up to 8 speakers). Unassign with U.
Keyboard Shortcuts:
- Next block: N or →
- Previous block: P or ←
- Jump to unassigned: Click on block in "Unassigned Blocks" list
- Split Block: Space - Opens split dialog to divide current segment
- Merge Blocks: Delete - Combine current block with next block
- Edit Content: E - Open text editor for current block
- Insert Empty Line: Enter - Add blank line for formatting
Access the symbols dialog with * or the Symbols button. Includes:
- Pauses: Micropause (.), short (-), medium (--), long (---)
- Breathing: Inhales (°h, °hh, °hhh) and exhales (h°, hh°, hhh°)
- Annotations: Comments ((())), actions (<<>>), overlaps ([ ])
- Measured Pauses: Custom timing with placement dialog
Load audio files via File → Import → Audio File. Supported formats: MP3, WAV, OGG, M4A, FLAC.
When importing audio, CapsGAT automatically searches the directory for matching subtitle files (SRT, JSON, TXT, TSV) and offers to import them.
Playback Controls:
- Play/Pause: End or click ⏯ button
- Rewind 5s: PgUp or click ⏪ button
- Fast Forward 5s: PgDn or click ⏩ button
- Jump to Time: Use the "Jump" button for precise navigation
- Adjust speed: Use the mouse wheel on the knob, the buttons (center button to reset speed) or the + and - keys to adjust playback speed
- Auto-sync to Audio: Automatically highlights the current transcript block during playback (Using audio controls [PgUp/PgDwn] instead of scrolling is recommended)
- Autopause: Automatically pauses audio when opening editing dialogs
- Progress Tracking: Visual progress bar with time display
Generate transcripts in these formats:
- HTML: Formatted with CSS styling, closer to the final, publication-ready transcript
- Plain Text: Simple text format for maximum copy-paste compatibility
- Subtitle file: An .SRT file containing time stamps and speaker diarization
Customize your export with these options:
- Include Timestamps: Add timecodes to transcript (when available)
- Include Diarization (Speaker Labels): Choose whether to include speaker labels (only applicable for subtitle exports)
- Project Title: Include project name as header
- Project Memo: Include project description
- Audio File Info: Include source audio file name
Use the preview feature to review your transcript formatting before exporting. The preview shows an approximation of how the final transcript will appear.