Architecture of zackees/transcribe-anything
zackees/transcribe-anything is 111 source files holding 800 symbols across 3 modules, written mainly in Python, Shell, Dockerfile. The largest module is tests, with 44 files, and 1 file is an entry point into the dependency graph.
How transcribe-anything fits together
The transcribe-anything command starts at src/transcribe_anything/__main__.py, which is the package's single entry point. Each Whisper backend runs inside its own isolated environment, so its dependencies do not conflict with the main install. The insane backend is implemented in src/transcribe_anything/insanely_fast_whisper.py, which handles the insanely-fast-whisper model path used by --device insane and --device insane-flash. src/transcribe_anything/server_app.py is a FastAPI application, and src/transcribe_anything/client.py is a central client file in the same package. ARCHITECTURE.md describes the design in more detail.
Largest modules
- tests: 44 files and 4,006 lines of code
- src: 41 files and 6,054 lines of code
- root: 17 files and 1,819 lines of code
- .github: 7 files and 246 lines of code
- .claude: 1 file and 9 lines of code
- examples: 1 file and 76 lines of code
Languages
By file count: Python 86 files, Shell 6 files, Dockerfile 1 file.
More on transcribe-anything: summary and FAQ, codebase overview, code health.
Related guide: Best dependency graph tools for monorepos