File Converter, PDF OCR and Audio Transcription (Whisper) in one Webapp for selfhosting.
100K+
Use filewizard:small for an Image without calibre, libreoffice and texlive. Use filewizard:lateest for a full image with every tool (also includes a tex build).
Here is the GitHub: https://github.com/LoredCast/filewizard
Run the image with docker-compose.yml:
version: "3.9"
services:
web:
image: loredcast/filewizard:0.4-latest
build:
context: .
# Set the target stage for the build. This is the crucial part.
target: full-final # Change to 'full-final' or 'small-final' as needed
#env_file:
# - .env
environment:
- LOCAL_ONLY=True # set to False to enable OIDC auth (requires configuration in settings.yml)
- SECRET_KEY= # if using auth
- UPLOADS_DIR=/app/uploads # directory structure INSIDE the container, not on host
- PROCESSED_DIR=/app/processed
- OMP_NUM_THREADS=1
- DOWNLOAD_KOKORO_ON_STARTUP=true
#- TRANSCRIPTION_DEVICE=cpu # change to 'cuda' if GPU available
#- TRANSCRIPTION_COMPUTE_TYPE=int8 # change to 'float16' or 'int8_float16' if GPU available
#- TRANSCRIPTION_DEVICE_INDEX=0 # change if multiple GPUs available
#user: "1000:1000"
ports:
- "6969:8000"
volumes:
# Mount local directories and files into the container for persistence
- ./config:/app/config # you can copy settings.default.yml into here as settings.yml
- ./uploads_data:/app/uploads
- ./processed_data:/app/processed
# For CUDA builds, enable GPU access (uncomment below)
#deploy:
# resources:
# reservations:
# devices:
# - driver: nvidia
# count: all
# capabilities: [gpu]
volumes:
uploads_data: {}
processed_data: {}
and
docker compose up -d
You can also add an environment file for further configuration:
.env
# Application Settings
LOCAL_ONLY=True
SECRET_KEY=test_key_for_dev
# File Paths (Development defaults)
UPLOADS_DIR=./uploads
PROCESSED_DIR=./processed
CHUNK_TMP_DIR=./uploads/tmp
Content type
Image
Digest
sha256:6ef0bc4d0…
Size
1.7 GB
Last updated
11 months ago
docker pull loredcast/filewizard