qbestd_box

All-in-one Docker container for deploying a Query-by-Example Spoken Term Detection system

Usage

1. Clone this GitHub repository

   git clone https://github.com/parledoct/qbestd_box.git
   cd qbestd_box

2. Set up environment:

   a. Use supplied docker container (recommended):

   # Launch parledoct/qbestd_box container with 
   # settings defiend in docker-compose.yml
   docker-compose run --rm search

   b. Or set up your local environment with required packages:

   pip install -r requirements.txt

3. Run a search using qbestd.py

   Run a search using the gos-kdl_queries collection (0375ff...) as queries and gos-kdl_corpus as test items (3e2d65...); for collection IDs see data/sqlite/collection_names.csv.

   python qbestd.py \
   	--progress \
   	--concurrent \
   	0375ff9c2d8555e88be65079d6c42df3 \
   	3e2d65d4486b2059ec9802a4221aacc0

   You can use python qbestd.py -h for usage instructions:

   usage: qbestd.py [-h] [-p] [-c] [-w MAX_WORKERS] query_id test_id
   
   example: python qbestd.py c4f0f58d1af2223da0519dc0496e7600 afeb2b96e36f1b38548959b3494a91e7
   
   positional arguments:
     query_id              Identifier for a query features file or a query collection.
     test_id               Identifier for a test item features file or a test item collection.
   
   optional arguments:
     -h, --help            show this help message and exit
     -p, --progress        show progress bar. (default: False)
     -c, --concurrent      run DTW searches concurrently. (default: False)
     -mw MAX_WORKERS, --max_workers MAX_WORKERS
                           if running concurrent jobs, maximum number of workers (None = use all available cores) (default: None)