Configuration
Important defaults live in shortificator/config.py.
| Constant |
Default |
Meaning |
WHISPER_MODEL_SIZE |
large-v3 |
faster-whisper model |
YUNET_MODEL_PATH |
models/face_detection_yunet_2023mar.onnx |
local face detector model |
DEFAULT_OUTPUT_FPS |
30 |
output FPS cap |
SHORT_MIN_SECS |
30 |
default minimum clip duration |
SHORT_MAX_SECS |
60 |
default maximum clip duration |
FACE_SMOOTH_WINDOW |
15 |
frames used to smooth tracking |
FACE_DETECT_EVERY |
3 |
detector interval in emitted frames |
FACE_DETECT_MAX_WIDTH |
960 |
max width for detection copy |
PIP_VISIBLE_THRESHOLD |
0.15 |
face/frame area threshold for PiP classification |
WORDS_PER_SUBTITLE_CHUNK |
5 |
static subtitle chunk size |
SUBTITLE_FONT_PX |
52 |
static subtitle font size |
Environment variables
| Variable |
Purpose |
OUTPUT_LANGUAGE |
default language for generated hook/reason text |
SUBTITLE_FONT_PATH |
override the TrueType font used for subtitles |