Suryansh Shakya’s Post

Tuning Word2Vec with Bayesian Optimization: Applied to Music Recommendations via #TowardsAI →

Can you use a AI music generator to create something with a symphonic, 1950s "hi-fi" vibe? That was the creative brief I gave myself: Create something with a lush string arrangement and rich chord changes — music you might expect to come from one of those magnificent hi-fi consoles. Something you might listen to after mixing a martini. Not the sort of music most people are creating with AI music tools. Here's my result, created using Udio. And as say again and again, I'm blown away by AI music platforms. In fact, I find myself more impressed than I am with AI tools that generate images and video clips. Maybe it's because I come from a family of musicians and play some instruments myself. Maybe it's because music is temporal: it unfolds over time, with structures and dynamic changes that engender emotion in a way that individual images and video clips don't convey. Maybe because music is so deeply rooted in our collective consciousness — by some theories, it's older than language itself. Finally, a production note (no pun!): You can't create a tune like this by simply typing a prompt and then sitting back and waiting for the results. What you'll hear below is the result of 27 separate generations. You build a song one segment at a time, choosing and combining the ones that sound best (some sound terrible) and that make the most musical sense (some are awful). You're composing alongside the AI. Have a listen and let me know what you think. https://lnkd.in/d3Pky7Fx #artificialintelligence #AImusic

Tuning Word2Vec with Bayesian Optimization: Applied to Music Recommendations via #TowardsAI → https://bit.ly/3Q5mY8T

Generative AI is a fascinating and also highly potential bearing topic. I am engaged with that for some time now, exploring it as well from the business side as from private usage. In the last week a lot of things happened: 🚀 OpenAI released GPT4o. A significant step in business as in private use. Faster, more capable in analytics, more human, better multimodality. AI comes even closer to everyone’s daily life with human like understanding and interaction. At least more of that. I might go into that in a later post. 🚀 Google integrates AI massively in all their products and Gemini 1.5. Immense capabilities to work with huge documents and much more. 🚀 Grok (within X) is available in Europe. I have access to it and am testing it 👍 🚀 Claude is available now in Europe as well. Tbh pretty disappointing for me in their Chatbot. Logic questionable (I did a few tests) and no web access. It’s quite eloquent and friendly, though. 🚀 Udio, the music generator, storms the scene with best quality…for now. It’s already available since 10th of April 24, but now the it really seems to take off. … and much more. For now, I want to share with you a small experiment.🧪 Can #Grok create songs? Yes! Lyrics, song structure and style suggestions. Here is what Grok came up for me. I just asked for Lyrics with verse, chorus etc. Grok chose everything else. Step by step developed in #Udio then. Please have a look 🎶🎧

More AI experimentation. Due to the length of the video, it had to be hosted off site. This video is an attempt to generate a John Williams style score, ala Raiders of the Lost Ark, via a generative AI service called CassetteAI. I generated about 16 different clips. In each case you will see the prompt. You will hear the music that was generated from the prompt. There aren't any visuals in this video outside of the prompt information. Although this service won't make composers obsolete -- yet -- it does allow you to download the midi files along with the finished mix. This means you could theoretically use any of these clips as the starting point for a film score or as a template. Pretty amazing stuff. https://lnkd.in/gCd3VzgB

One of my works that I have completely generated in multiple neural networks: https://lnkd.in/eKRfEcpS

The analogy between new music creation and content generated from GenAI is noteworthy. Just like we have a set of notes permutating in different ways, resulting in new music, the same way content created by GenAI is permutating existing words in a new way, that is, new content. https://lnkd.in/gCcmnapa

Suno V3.5 test =) `Final moments`: https://lnkd.in/dqFD7GHm Wrote only 1st verse and chorus rest was finished by GPT: https://lnkd.in/dSdNpBui

This music clip was made with an AI text prompt. Listen to the first 20 seconds. Now will AI have an impact? #ai #genai Paul W. Kevin Rank, MBA https://lnkd.in/gQbXhAVA

🎵 Excited to share my latest project: an Emotion Recognition-Based Music Recommendation System! Leveraging Convolutional Neural Networks, it detects emotions through facial expressions captured via webcam. Utilizing Django, PostgreSQL, and Python, it categorizes songs based on emotion, offering personalized playlists. Exciting journey merging deep learning and music! #AI #MusicRecommendation #EmotionRecognition #DeepLearning https://lnkd.in/gX_NzMcz