Songscription, a fledgling company, has launched an innovative AI-driven solution that revolutionizes the world of music transcription. Within minutes, this groundbreaking technology can transform an audio file of a song into sheet music, catering to both professional and amateur musicians. Andrew Carlins, the CEO of Songscription and a Stanford MBA/MA in Education student, envisions a future where music educators in remote areas can effortlessly access sheet music tailored to their students' instruments and skill levels.
At its inception, Songscription excels in transcribing music for various instruments, with the piano model being the most accurate. The company aspires to expand its capabilities, including the addition of guitar tabs and full band arrangements. This tool is particularly beneficial for musicians who wish to bypass manual transcription, as well as for those unfamiliar with sheet music, as Songscription also generates piano rolls, offering a digital visualization of the music on a virtual piano.
An intriguing feature of Songscription is the ability to transcribe music directly from YouTube links, although users must confirm they hold the rights to transcribe the uploaded files. This raises questions about the ethical and legal implications of using such technology, especially with copyrighted material. Carlins acknowledges the gray area surrounding the use of tech-enabled platforms but maintains that since music learners are allowed to listen to, transcribe, and perform songs at home without charging for the performance, Songscription's role as an augmented music notation software seems to be within legal boundaries.
The legality of creative AI tools is a contentious issue, with recent court rulings often favoring tech companies. However, Songscription does not generate new AI music; instead, it expedites the process of creating guitar tabs or sheet music for musicians. Carlins emphasizes that the platform's editing capabilities position it as a tool to accelerate transcription, not replace the creative process.
The AI model at the core of Songscription is rooted in a paper co-authored by co-founder Tim Beyer and researcher Angela Dai. To gather the necessary training data, Songscription collaborates with musicians willing to share or sell their performances and sheet music. In addition to using public domain sheet music, the company relies heavily on synthetic data, converting sheet music into audio and modifying files to mimic real-world scenarios, including background noise and reverb.
In just seven months since its founding, Songscription has secured pre-seed funding from Reach Capital and is set to join Stanford's StartX accelerator program. This marks a significant step forward in the company's journey to transform the way musicians create and engage with sheet music.