LyriGenesis is an Introduction to Artificial Intelligence (AI) Final Project. It is an AI model that takes a word or phrase from a user and uses it to generate song lyrics whose length is also based on another input by the user.
Music is a universal form of expression that transcends language and cultural barriers. Music lyrics are an important component of music, as they convey the emotions, thoughts, and stories of the artists to the listeners. However, writing music lyrics can be challenging, especially for novice or aspiring musicians who may lack the skills, experience, or inspiration to create original and engaging lyrics. Artificial Intelligence (AI) has been applied to various domains and applications, such as image recognition, speech synthesis, game playing, and self-driving cars. One of the emerging applications of AI is to generate music lyrics using techniques such as deep learning, natural language generation, and text analysis. The project is significant and has a social impact in several ways. First, it can provide a useful tool for musicians who want to generate song lyrics or get inspiration from AI-generated lyrics. Second, it can enhance the creativity and diversity of music lyrics by introducing new words, phrases, rhymes, and styles that human lyricists may not commonly use. Third, it can contribute to advancing AI research and development by exploring the challenges and opportunities of applying AI to a creative domain such as music.
- This project used a music lyrics dataset downloaded from the web to train and test the model.
- Only songs in the English language from the dataset were used in order to scale down the data and scope.
- Python on Google Colaboratory was used in creating, testing and training our model.
- The data was pre-processed to remove null values.
- Tokenizer was also used to break down the words into smaller units called tokens to help the model easily understand and process the data.
- The langdetect library was used to select the songs whose lyrics were in English.
- The Long Short-Term Memory (LSTM) machine learning technique was used to train the model.
The model was deployed on the web using Streamlit. Attached is a link to a video demonstration of how the deployed model operates: https://youtu.be/L1fOmpeHXqQ
In conclusion, our project will present a novel AI-based system for generating music lyrics for a given genre, mood, and theme. The system builds on several aspects of the technical and theoretical knowledge gained in the class so far, including neural networks, natural language processing, and text generation. The system has been evaluated on a dataset of music lyrics, and the results show that the system is able to generate lyrics that are diverse, creative, and consistent with the style of the input music. The system has the potential to be used by songwriters, composers, and music producers to generate new ideas for music lyrics. It can also be used to create personalized music experiences for listeners. The project is also significant and has a social impact in several ways. First, it can provide a useful tool for musicians who want to generate song lyrics or get inspiration from AI-generated lyrics. Second, it can enhance the creativity and diversity of music lyrics by introducing new words, phrases, rhymes, and styles that human lyricists may not commonly use. Third, it can contribute to advancing AI research and development by exploring the challenges and opportunities of applying AI to a creative domain such as music.