Content
@
0 reply
0 recast
0 reaction
joão
@jaywooow.eth
"Spectral Dreams" is cooking and will drop later this week 🔥 This gen art edition celebrates a "audio mode" feature release on DecentAI. The token will give holders full access for 30 days. It's a fully interactive 3D scene showing the beauty of speech. The process has been extremely fun. More details below 👇
4 replies
1 recast
10 reactions
joão
@jaywooow.eth
Step 1: Generate Audio 🔊 I had ChatGPT generate a few short poems about what it's like to be stuck in a machine then synthesized the audio using elevenlabs. I took the audio and generated the spectrogram data offline. Spectrograms are the core inputs to most speech-related AI models.
1 reply
0 recast
1 reaction
joão
@jaywooow.eth
Step 2: Render Spectrogram These spectrograms are usual because they are rendered radially ⭕ The intensity is mapped to a colormap and a shader I wrote that creates the textured effect. It's rendered in a 3D scene with Three.js which means it's a fully interactive scene with zoom and camera controls.
1 reply
0 recast
1 reaction
joão
@jaywooow.eth
Step 3: Mint Each mint is unique and uses onchain data to generate randomness (as a seed). Thanks again @highlight for the amazing workflow. Tokens unlock access to a decentralized alternative to ChatGPT called /decent-ai. 500 will be available. Wave 1 holders will get an airdrop. Learn more: decentai.xyz
2 replies
0 recast
2 reactions
Ashira 🎩🫂🍖🍔💃🏼
@ashira
Looks dope!!
0 reply
0 recast
1 reaction
Marissa
@marissaposner
This is so cool! Do you ever think there is a world where spectrograms could be generated on chain?
0 reply
0 recast
0 reaction
