Songscription Launches AI-Powered “Shazam for Sheet Music” Tool

A recently launched company called Songscription debuted last week, providing AI models that automate the process of music transcription by converting audio files of songs into sheet music within minutes. The service adopts a freemium model, catering to both professional musicians and hobbyists.

“Our mission is to elevate the joy of making music,” said Andrew Carlins, CEO of Songscription and a student in Stanford’s MBA/MA in Education program, in an interview with TechCrunch. “We envision a future where a high school band director in rural Nebraska can easily access sheet music for the songs their students want, with arrangements customized for the specific instruments and skill levels of each student.”

At launch, Songscription is capable of transcribing music for a variety of instruments, with its piano model being the most reliable. Looking ahead, the company plans to roll out more transcription options (like guitar tabs) and arrangements for entire bands instead of just one instrument.

This service can greatly benefit musicians who record their original compositions and then upload them to create sheet music, effectively removing the need for manual transcription. For users who don’t excel at reading sheet music, Songscription also provides a piano roll, which is a digital representation of the music as played on a virtual piano.

Users can also automate music transcription directly from YouTube links. When submitting a file, users must check a box confirming their rights to transcribe the content, which raises the possibility of acquiring free sheet music for copyrighted materials.

“For music learners… since you can listen to a song, transcribe the notes by ear, and play it on your home piano (as long as you don’t charge for performances), it’s somewhat unclear whether using a tech-enabled platform for help crosses any legal boundaries. Nonetheless, we acknowledge that this area is evolving and our application may dwell in a legal gray area,” Carlins remarked.

The legal implications of our interactions with these creative AI tools are still up for debate, yet recent court rulings seem to favor tech companies over artists. Nevertheless, Songscription is not producing original AI-generated music; it merely offers a tool that streamlines the creation of individual guitar tabs or sheet music.

“Our platform lets users edit their scores, positioning us as an advanced music notation software that facilitates quicker transcription,” Carlins noted.

The foundation of Songscription’s AI model is based on research by co-founder Tim Beyer, published in collaboration with researcher Angela Dai.

To acquire the training data necessary for its AI model, Songscription partners with musicians who agree to share or sell their piano performances and sheet music. The company also utilizes public domain sheet music; however, according to Carlins, most training data is synthetic. In these cases, Songscription converts the sheet music into audio and then adjusts the files to mimic real-world conditions, such as background noise or reverb.

Just seven months after its inception, Songscription has successfully secured pre-seed funding from Reach Capital and will participate in Stanford’s StartX accelerator.