Technologies
Deep Learning
Many of the piano accompaniments attached to Ladukhin's melodies are based on scores created using deep learning techniques.The AI, trained by Bach's chorale, is used to recognize Ladukhin's melody as a soprano part, and the AI automatically creates the appropriate alto, tenor and bass parts. For the automatic creation of the parts, "BachBot", a research project at the University of Cambridge was used.
Singing Voice Synthesis
The vocal audios used in this app are generated using "Sinsy", a vocal synthesizer developed by the Nagoya Institute of Technology in a research project. Sinsy is a system that uses Hidden Markov Model to synthesize a singing voice. If you upload MusixXML with lyrics, you can download the synthesized vocal audio file.
