GrantL10 / sp2si-code

Contains code for our work on speech to singing conversion (ICASSP 2020)

Speech-to-Singing Conversion

• This repo is for the paper: "Speech-to-Singing Conversion in an Encoder-Decoder Framework" by Jayneel Parekh, Preeti Rao, Yi-Hsuan Yang in ICASSP 2020.

• Link for the project webpage

• If you are planning to use this code already, proceed with caution! Documentation is being updated currently.

Setup

You can setup a new conda environment with the environment.yml file (recommended). You can start with miniconda installation if you are completely unfamiliar with anaconda

conda env create -f environment.yml

Data

You will need to download the dataset (NUS-48E) and the models weigths to run the code.

• The NUS-48E dataset can be downloaded from this link. The downloaded dataset (folder named 'NUS_48E') should be placed in this repo.

• The model weights can be downloaded from here. You should place the complete folder (named 'models') in the 'output' folder of this repo.

Usage (Being updated)

The first time you run the code, it will also organize the audio from NUS-48E in a dictionary. This can take up to 5-10 minutes.

You can currently

1. Compute LSD for different models on random samples generated from NUS-48E dataset (with the function eval_sys()).
2. Compute random predictions for multiple models on the NUS-48E data (function random_pred()).

python evaluation_sp2si.py

To be added

1. Pitch extraction related stuff
2. General function to run the model on any speech file and singing file Need to rewrite these parts as I lost them when my hard drive crashed. Will add within a few weeks