SongGen: Framework for Controllable AI Song Generation through Interactive Songwriting and Artist Emulation

We propose SongGen, an AI-based song-writing and song co-creation framework. Building upon existing AI tools like Suno.ai, SongGen features a chat interface with a trained AI songwriter assistant, emulating the traditional back-and-forth of human collaboration. The system offers enhanced capabilities for greater control over the songwriting process, including concept ideation, lyric generation and editing, real-time song generation, and granular instrumental specification. Comparative evaluations demonstrate SongGen’s superiority in key metrics such as steerability, expressiveness, personalization, and user satisfaction. We also present an extension of the SongGen framework for artist emulation and on-demand song generation. Future development aims to incorporate voice-based interaction and real-time voice conversion, enabling music artists to guide fans in creating personalized songs.

Name thesis:degree_name

Master

Department dc:contributor.department

Massachusetts Institute of Technology. Department of Electrical Engineering and Computer Science

Grantor dc:publisher

Massachusetts Institute of Technology

Year dc:date.issued

2024

Author dc:creator

• Arora, Ajay

Advisors dc:contributor.advisor

• Egozy, Eran
• Jaco, Wasalu

Statement dc:rights

• In Copyright - Educational Use Permitted
• Copyright retained by author(s)

Licence dc:rights.uri

• https://rightsstatements.org/page/InC-EDU/1.0/

Handle dc:identifier.uri

https://hdl.handle.net/1721.1/157249

OAI identifier oai:identifier

oai:dspace.mit.edu:1721.1/157249