OpenAI — Jukebox [Attempt #3]

After preparing my laptop to run this model locally with the appropriate library installs, I ran into another blockade. The Google library installed onto the machine but loaded without end once its cell was engaged in the .pynb file.

After hours of waiting by the computer, leaving the computer to get food, watching a Netflix movie, studying up on blockchain, and returning to the computer, I accepted that it did not work. Sadly, it was time to troubleshoot. One word echoed in my mind for a day and that word is debugging.

I started the process of debugging by deactivating jupyter notebook, checking for errors in Anaconda’s command prompt, and reinitializing jupyter notebook. Google’s library import yielded the same results which were less than ideal.

I may have to go back to Google Colab and purchase more storage to make this work or train my model with OpenAI’s sample library. Regardless, I have spent a good amount of time researching about Jukebox and hearing majority of the results from different users on the internet who have used this model in conjunction with songs from Jukebox’s sample library. I also heard sounds of those who imported custom music into the model. Honestly, all the results were lackluster but this is a major step towards having generated music play a role in the music industry. I can see this shifting the music industry, giving voices to those behind their computers in music. It may take some time but it will come.

I will have another hand at this project. After this next attempt, I will be moving on to another project.