Real-Time Lyric Generation Technology is Being Developed by Researchers to Inspire Songwriting

Real-Time Lyric Generation Technology is Being Developed by Researchers to Inspire Songwriting

Musicians may use technologies created by Waterloo researchers to discover inspiration and new creative avenues for their songwriting. Members of the University’s Natural Language Processing Lab developed LyricJam, a real-time system that employs artificial intelligence (AI) to produce lyric lines for live instrumental music.

For several years, the lab, directed by Olga Vechtomova, a Waterloo Engineering professor with a cross-appointment in Computer Science, has been investigating innovative AI applications. The lab’s first project resulted in the development of a system that learns artists’ musical expressions and creates lyrics in their manner.

Vechtomova and Waterloo graduate students Gaurav Sahu and Dhruv Kumar have created technology that synthesizes lyrics reflecting the atmosphere and emotions portrayed by live music using different characteristics of music like as chord progressions, pace, and instrumentation. Get comfortable with journaling and utilizing your senses when learning how to compose lyrics for a song. Taste, touch, sight, sound, smell, and movement are adjectives that can assist you immerse your listener in a brief moment’s experience.

A little moment is a slice of life, a setting in which your song takes place. While a performer or band is playing instrumental music, the system gets raw audio samples, which the neural network analyses to produce new lyric lines. The artists can then utilize the lines to write their own lyrics for their songs.

“The purpose of the system is not to write a song for the artist,” Vechtomova explains. “Instead, we want to help artists realize their own creativity. The technology creates poetry lines with fresh metaphors and phrases, which might take the artists down new creative paths they hadn’t considered before.”

The researchers’ neural network learns which poetic themes, phrases, and stylistic techniques are linked to various components of music collected in each audio clip. The researchers discovered that lyrics created for ambient music differ significantly from those created for cheerful music.

Don’t feel obligated to compose songs in the same manner as someone else because everyone has their own unique style. It’s perfectly OK to approach songwriting in a unique way! Some people write spontaneously from their imaginations, while others write with a definite goal in mind. While music has many laws and traditions, it is ultimately a creative endeavor, which means that the most essential thing is that it reflects you. A user study was done by the research team, which invited artists to utilize the system while playing live instruments.

“One unexpected finding was that participants felt encouraged by the generated lines to improvise,” Vechtomova said. “The lines, for example, prompted artists to restructure chords and take their improvisation in a different way than they had planned. The lines were also utilized by certain musicians to see if their improvisation had the intended emotional effect.”

Another research finding focused on the co-creative component of the event. Participants said they saw the system as a nonjudgmental jamming buddy who encouraged them to play their instruments even if they weren’t consciously attempting to create songs.

Over 1,500 individuals have tried LyricJam since it launched in June of this year. The team’s research, to be presented at the International Conference on Computations Creativity this September, has been pre-published on arXiv. Musicians interested in trying out LyricJam can access it at