Generate embeddings with bug-fixes

[auroracramer]

US8K

• L3 embedding / linear model / music set
• L3 embedding / melspec1 / music set
• L3 embedding / melspec2 / music set
• L3 embedding / linear model / env set
• L3 embedding / melspec1 / env set
• L3 embedding / melspec2 / env set

ESC-50

• L3 embedding / linear model / music set
• L3 embedding / melspec1 / music set
• L3 embedding / melspec2 / music set
• L3 embedding / linear model / env set
• L3 embedding / melspec1 / env set
• L3 embedding / melspec2 / env set

DCASE 2013

• L3 embedding / linear model / music set
• L3 embedding / melspec1 / music set
• L3 embedding / melspec2 / music set
• L3 embedding / linear model / env set
• L3 embedding / melspec1 / env set
• L3 embedding / melspec2 / env set

Instructions

To generate samples (for L3 embedding models), make edit or make a copy of jobs/generate_embedding_samples_array_<dataset name>.sbatch In there, you'll want to make the following changes:

0. Make sure that the embedding model path follows the convention <output_dir>/embedding/<music or environmental>/<l3 model type string>/<timestamp>/model_best_valid_accuracy.h5. The code assumes this convention to be followed.
1. Change the email to your email
2. Change the anaconda environment name
3. Change SRCDIR to whatever directory the l3embedding repository is in
4. Change L3_MODEL_PATH to the path of the .h5 file of the model you want to use.
5. Change L3_POOLING_TYPE to original.
6. Change OUTPUT_DIR to your general experiment output directory. A subdirectory called features will be created here for the embedding output.
7. Add any additional command line arguments to the script call. Take a look at 05_generate_embedding_samples.py for the options. (shouldn't need to do for this though)
8. Run sbatch --array=1-<num folds> generate_embedding_samples_array_<dataset>.sbatch

• is 10 for US8K, 5 for ESC-50, and 2 for DCASE