Music Composition with Deep Learning (Project 243)

[robertogl]

Contribute to the discussion by asking and/or answering questions, commenting, or sharing your ideas for solutions to project 243

[luciry]

Hello, my name is Luca, I study Strategic Engineering in the University of Genova. I'm interested in the project and I will work on it for my exam of Machine Learning and Data Analysis

[cps13]

Hi Luca, good luck with the project. I'm excited to see how it progresses!

[luciry]

Thank you very much! 😄

[Akshada07pute]

Can we do this project along with some hardware like microcontroller and speaker

[cps13]

That sounds like a fun idea! You are welcome to expand the project description however you would like to.

[Akshada07pute]

As we get the output in the musical sound how can we make the output coming from the speaker which is interfaced with microcontroller can you please suggest some idea

[cps13]

This will depend on what exactly the output of your model is. In the project description, we recommend outputting a MIDI sequence that can then be played in a 3rd party tool, or you can synthesize audio with the Audio Toolbox. That could be a good place to start:

• Play the resultant sequence, either by writing a MIDI file and playing it outside of MATLAB, or synthesizing audio with the Audio Toolbox™.

[Arnav2Prasad]

Hi, I'm Arnav Prasad, a BTech CSE student at COEP Technological University, Pune, India.

Is the problem statement : "Develop a solution where a user inputs an image of musical notation, and the system identifies the notes and plays the corresponding tune" aligns in to yours problem statement?

[cps13]

It is a little different, but that sounds like a great project, too!

[Arnav2Prasad]

Okay, I’ll definitely make some progress in this project and would love the opportunity to present it to you. Looking forward to our continued discussion!

[TanV-04]

Hi, I'm Tanvi! I'm a third-year CSE undergrad at SIT, Pune, India.

My teammates and I completed this project as part of our MATLAB elective in college, developing a basic yet efficient version based on the concepts and techniques we learned in class.

Here's the GitHub repository link: https://github.com/TanV-04/ESMT_Project/tree/master

We’d love to hear your feedback and suggestions to help us improve further!

[aksh-sush]

Hey! I'm really excited to check it out. However, when I cloned your repo, it didn’t show me any files. Did I maybe copy the wrong branch?

[TanV-04]

Try downloading the zip, if cloning doesn't work.

[sohamkorade]

From the project description:

Use the Deep Learning Toolbox™ and the Audio Toolbox™ to create a generative model that can compose and play back a melody.

Is using the Deep Learning Toolbox™ or Audio Toolbox™ compulsory for the project? Can I use other libraries such as PyTorch for the deep learning model?

[robertogl]

using MATLAB is a requirement, unless you need features not available in MATLAB. In that case you can still use MATLAB in combination with other products.

[aksh-sush]

Hi, I’m a computer engineering student from India, and I’m really interested in visualizing music and much more. I’ve worked on analyzing and synthesizing Indian music, but since I’m new to MATLAB, everything still feels pretty raw and a bit intimidating. I’m still figuring out how to build a proper toolkit. So out of curiousity—what do you all use, and for what purposes? Are there any plugins or tools you’d consider absolute must-haves?

[TanV-04]

Some toolboxes you can use are:

• Signal Processing
• Audio Toolbox
• Image Processing Toolbox
• Spectrogram
• Musical Toolbox

These should help you get started.

[aksh-sush]

I couldn’t find the Spectrogram and Musical Toolbox tools. Anyway, instead of the Musical Toolbox, can I use the Signal Processing Toolbox? I’ll have to refine the audio and feed it into some ML, then work with it key by key.

[TanV-04]

Yeah, that works too. Its all about experimentation.