Show HN: Tweak your chord progressions for practice or composition https://ift.tt/wJdixVX

Show HN: Tweak your chord progressions for practice or composition I'm excited to share with HN a new pet project I've been working on to explore how AI can create and explain harmonic chord progressions - Chord Variations. The project uses GPT-4 to generate unique and interesting chord combinations based on user input chord progression. It's akin to having a virtual musical assistant that can help non-musicians and musicians alike explore and create harmonious sound without needing any prior knowledge of music theory. The generated chord suggestions maintain a similar vibe to the user input. Alternative chord progression includes extended chords, chord substitutions, unique passing chords, and more. Additionally, musical theory explanations within the tool is helpful for users not just to create music, but understand the underlying structure it's built on a bit better. These chord progression suggestions can be used for practice or composition. As a musician myself, I am having quite a lot of fun playing around with it. One of the things I'm proud of is how the application really dips into music theory. It includes nuanced aspects such as dominant chords, secondary dominant chords, and the famous 2-5-1 chord progressions. That said, the development journey was full of lessons. Dealing with the latency of the GPT-4 API was particularly challenging. I used a Celery based queue system + client polling to manage the delay between request and response (from OpenAI API). Additionally, to keep the chord names consistent, I used a combination of prompting and regex. There are still some bugs that need to be squashed but overall I am pretty happy with the results. I'd love to hear your thoughts and feedback on the project. Also, if you're curious about anything, I am happy to delve into the details in comments. Feel free to take Chord Variations for a spin here: https://ift.tt/sUyrCqA Looking forward to some interesting discussions! https://ift.tt/sUyrCqA June 28, 2023 at 12:47AM

Comments