Music generation with deep learning techniques
This report demonstrated the use of conditioning inputs, together with an appropriate model architecture, to improve the structure of generated music through deep learning. Existing challenges to generate music using deep learning, in particular structure, were reviewed. The use of bar counter, occu...
| Main Author: | |
|---|---|
| Other Authors: | |
| Format: | Final Year Project (FYP) |
| Language: | English |
| Published: |
Nanyang Technological University
2021
|
| Subjects: | |
| Online Access: | https://hdl.handle.net/10356/153284 |
| _version_ | 1826130117160599552 |
|---|---|
| author | Lee, Daniel Yu Sheng |
| author2 | Alexei Sourin |
| author_facet | Alexei Sourin Lee, Daniel Yu Sheng |
| author_sort | Lee, Daniel Yu Sheng |
| collection | NTU |
| description | This report demonstrated the use of conditioning inputs, together with an appropriate model architecture, to improve the structure of generated music through deep learning. Existing challenges to generate music using deep learning, in particular structure, were reviewed. The use of bar counter, occurrence of repeated motifs, and form of a piece as conditioning inputs were hypothesized to capture long-term structure of music. Then, the proposed model was designed using Bidirectional Long Short-Term Memory (Bi-LSTM) and attention layers to take in the conditioning inputs. To evaluate performance of the proposed model, quantitative analysis was done on the proposed model, the same model without conditioning inputs, and a baseline LSTM model. Following which, a user study was conducted to compare music samples generated by the 3 models. Evaluation results verified that by utilising the 3 conditioning inputs, the proposed model generated more pleasant-sounding and structurally coherent music. |
| first_indexed | 2024-10-01T07:51:34Z |
| format | Final Year Project (FYP) |
| id | ntu-10356/153284 |
| institution | Nanyang Technological University |
| language | English |
| last_indexed | 2024-10-01T07:51:34Z |
| publishDate | 2021 |
| publisher | Nanyang Technological University |
| record_format | dspace |
| spelling | ntu-10356/1532842021-11-16T00:19:09Z Music generation with deep learning techniques Lee, Daniel Yu Sheng Alexei Sourin School of Computer Science and Engineering assourin@ntu.edu.sg Engineering::Computer science and engineering::Computing methodologies::Artificial intelligence This report demonstrated the use of conditioning inputs, together with an appropriate model architecture, to improve the structure of generated music through deep learning. Existing challenges to generate music using deep learning, in particular structure, were reviewed. The use of bar counter, occurrence of repeated motifs, and form of a piece as conditioning inputs were hypothesized to capture long-term structure of music. Then, the proposed model was designed using Bidirectional Long Short-Term Memory (Bi-LSTM) and attention layers to take in the conditioning inputs. To evaluate performance of the proposed model, quantitative analysis was done on the proposed model, the same model without conditioning inputs, and a baseline LSTM model. Following which, a user study was conducted to compare music samples generated by the 3 models. Evaluation results verified that by utilising the 3 conditioning inputs, the proposed model generated more pleasant-sounding and structurally coherent music. Bachelor of Engineering (Computer Science) 2021-11-16T00:19:09Z 2021-11-16T00:19:09Z 2021 Final Year Project (FYP) Lee, D. Y. S. (2021). Music generation with deep learning techniques. Final Year Project (FYP), Nanyang Technological University, Singapore. https://hdl.handle.net/10356/153284 https://hdl.handle.net/10356/153284 en application/pdf Nanyang Technological University |
| spellingShingle | Engineering::Computer science and engineering::Computing methodologies::Artificial intelligence Lee, Daniel Yu Sheng Music generation with deep learning techniques |
| title | Music generation with deep learning techniques |
| title_full | Music generation with deep learning techniques |
| title_fullStr | Music generation with deep learning techniques |
| title_full_unstemmed | Music generation with deep learning techniques |
| title_short | Music generation with deep learning techniques |
| title_sort | music generation with deep learning techniques |
| topic | Engineering::Computer science and engineering::Computing methodologies::Artificial intelligence |
| url | https://hdl.handle.net/10356/153284 |
| work_keys_str_mv | AT leedanielyusheng musicgenerationwithdeeplearningtechniques |