This is a minimum viable product for the music-for-the-moment application. This application queries speech-to-text APIs on audio data and then runs IBM Watson's text tone analyzer to search a code base of music and play something relevant to the tone of the audio. Users create playlists on Youtube the application selects music from them. We use Genius APIs for song lyrics and run tone and relevancy scoring to select the most relevant songs in the Youtube playlist from lyrics and song titles and play them through youtube. We can also interface with downloaded music in local files.
Here are some tested conversations we used for our demo on a pre-created youtube playlist to test our application if you like: