ENS Audiobound

Screenshots

Problem Statement

Unleash creativity with AI, for a soundful web 3.0. Audiobound builds AI-native workflows to power faster, easier, and better sound creation. ENS is the unique identity for most web3 users. By linking with the ENS domain, Audiobound creates another layer of users' digital identity and enables users to sound like anyone on the web3. So grab your ENS as seed NFT, plant it in the soil of AI and hear the magic voice.DID and pfp NFTs provide people the freedom to look however people want. For example, my digital identity is 1201.eth with a cat avatar. We believe that voices need the same freedom. A person's voice is uniquely linked to their identity. People will want a complete, new virtual identity to represent themselves, and people's voices are crucial. A particular minority of people like the sound in real life and will use it in Web 3.0, but most people would like to have the freedom of choice.Audiobound provides access to a unique AI-powered voice font that essentially becomes your voice identity in web3. Audiobound, built on top of ENS, empowers customized and diverse audio experiences in the ENS ecosystem. In addition to minting it as an NFT, we also make a protocol that allows ENS users to link any audio NFT with the ENS domain through text-record. Users then can integrate audio NFTs into the vast ENS ecosystem and take advantage of it in a variety of scenarios:Play audio/music as part of their web3 profile on social-fi platforms;Use audible identity as personalized BGM when playing metaverse games;Play your audible identity in any wallets that support ENS."AI to the left, Crypto to the right." Our project provides generative AI to free your creativity and power your engagement. We can build voice models from your YouTube, podcast, or recordings. We can also generate an AI voice that can be used permissionless. You can now create customized audio content for your fans, like reading the names of your most loyal fans and reading particular messages for them. It can be as simple as a "Good morning, Suda" with the voice of Vitalik or any other creator. Easy, you can always create with AI without any prerequisite in your skillset.

Solution

ENS:Read all users' ENS domains from the walletGenerate AI audio NFT with visual art where rarity comes from the hash of domainCreate a new text record "audio" with the linkage to users' audio identity NFTIPFS:Host over 10,000 NFTs on IPFSFast query of data via decentralized gatewayThe Graph:Read users' ENS domains through ENS subgraphUser-Facing contract and subgraphAI pipeline:Download 24 hours of Vitalik's tech talks and split them into sentences, each as an individual audio clipImplement a basic volume normalize to process all audio clipsUse a 3rd party transcription service to annotate all audio clipsBuild Annotation Service and conduct annotionTraining an open source TTS model with prepared training dataFrontend:Design and development of user demo websiteTesting:User end-to-end testing