Multimodal learning using IMEOCAP dataset.

Refer to config.ini to configure the paths of audio and text dataset. The config file can also be used to add hyperparameters such as learning rate and dropout.

Clone the code, configure the config.ini file. Run python emotion_utils\main.py. The main.py file also takes some arguments. For more details run

python emotion_utils\main.py --help

Use the slurm job (lightning_job.sh) as a template to run on HPC. You can also check the usage of code in the slurm job file. The slurm job can be run using the following command:

sbatch lightning_job.sh

The dataset is accessible here:https://sail.usc.edu/iemocap/.