Project Context
This was a client project where I was tasked with creating a high-fidelity mockup for an AI music app. The app needed a user flow allowing incremental voice training with an AI model, enabling users to progressively train their voice rather than all at once, and then use the trained voice to generate songs. Additional requirements included a monetization system, a music library, and a trained voice library.
The goal was to design an app interface reflecting the identity of an AI startup. The user flow had to be clear and consistent, support incremental voice training, and provide straightforward navigation for accessing generated songs. Additionally, an accompanying landing page design was required.
Research
The process began with research into how current AI models, specifically Apple's personal voice feature, learn people's voices. It was important to understand this technology thoroughly to create an appropriate user flow. In depth benchmarking informed the approach to project. Companies such as Songlorious and Songfinch were used to evaluate how existing services function, and how to differentiate on the market. This stage of development proved crucial from a product development perspective.
Wireframes, mood boards, and mockups were developed and presented to the client, alongside explanations of existing AI voice training technologies. After client approval, I expanded the concept through detailed UI mockups.
Development
UI Style
Various UI styles were explored, including neomorphism, flat design, and skeuomorphism, before selecting glass-morphism to align the app design with existing AI startup trends.
Navigation
Navigation was a critical component of the app, given the multiple functionalities required. The most significant challenge was developing a voice training section engaging enough for users to comfortably speak for five minutes with provided prompts to effectively train the AI model.
User Feedback
Creating a user-friendly system to clearly guide users through voice training was important. High-fidelity mockups created in Figma allowed for effective user testing, ensuring clarity in training requirements and model accuracy feedback.
Trust
Establishing user trust regarding data privacy was a priority. I pushed for the development of a system that stored the collected data locally, enhancing data protection. This was important because of people are becoming more aware of AI systems using stolen data. This approach was important for user adoption, and was a key consideration throughout the design process.
Final Design
The final app design successfully fulfilled the project requirements, featuring detailed mockups for:
Login / Sign-Up: An intuitive onboarding experience for new and existing users.
Monetization System: A credit-based system for song generation.
Song Generation Page: Users could generate songs using semantic descriptions and their trained voice.
Music Library: Storage for all AI-generated songs.
Profile Page: Simple management of personal details and credit tracking.
Voice Library: Repository of trained voices, including feedback on voice training accuracy.
Voice Training User Flow: Incremental AI voice training with progress tracking.
An accompanying landing page was also developed: https://promusic.ai/
The finalized design was handed off to a developer to create an MVP, enabling market testing and assessment of investor interest. This project significantly expanded my skills in Figma, UX/UI design, and provided valuable experience in professional freelance client collaboration. The project concluded mutually due to external factors inhibiting further development.
More Photos of the final design
Please contact me for more details on this project
