Songscription launches an AI-powered ‘Shazam for sheet music’

Songscription, a budding tech startup, recently made headlines with the introduction of an innovative AI-driven tool designed to transform audio files into sheet music in mere minutes. This service, which operates on a freemium basis, caters to both seasoned musicians and enthusiastic amateurs alike. Andrew Carlins, the CEO of Songscription and a current MBA/MA student at Stanford University, shared his vision with TechCrunch: "We aim to enhance the music-playing experience for everyone. Imagine a high school band director in rural Nebraska being able to obtain tailored sheet music for their students, arranged specifically for the instruments in the band and at the individual skill level of each player." Currently, Songscription's technology can transcribe for various instruments, with the piano model showing the highest reliability. Looking ahead, the company plans to expand its offerings to include different transcription formats, such as guitar tabs, and arrangements suitable for full bands. This tool could significantly benefit musicians who want to record their compositions and quickly convert them into sheet music, eliminating the tedious manual transcription process. Additionally, for those unfamiliar with reading sheet music, Songscription can generate a piano roll, providing a digital visualization of the music played on a virtual keyboard. Users can even automate transcription directly from YouTube links. While uploading files requires users to confirm their rights to the material, there are concerns about the ease with which someone could access free sheet music for copyrighted songs. Carlins acknowledged the gray legal areas surrounding the use of such tech-enabled platforms, stating, "For music learners, as long as you can listen to a song, jot down notes by ear, and play it at home without charging for performances, it's unclear if using our platform crosses any legal lines." Despite ongoing debates regarding the legality of AI tools in creative fields, recent court rulings appear to favor technology companies. Songscription, however, does not generate new music but rather serves as an efficient tool for composers looking to accelerate their music transcription processes. "Our platform enables users to edit scores, positioning us as an augmented music notation software that streamlines transcription," Carlins explained. The foundation of Songscription's AI model comes from research conducted by co-founder Tim Beyer in collaboration with researcher Angela Dai. To develop their AI technology, the company collaborates with musicians willing to share their piano performances and sheet music while also utilizing public domain materials. A significant portion of their training data is synthetic, with Songscription converting sheet music into audio and simulating real-world conditions like background noise or reverb. Just seven months after its inception, Songscription has successfully secured pre-seed funding from Reach Capital and is set to join Stanford's StartX accelerator program.