Final Project Ideation

1. MindMeld - AI Emotional Companion

• Gentle daily check-in that helps anyone notice patterns in their emotional state without feeling clinical.
• User speaks or types for 30-60 seconds, the app listens for tone/word cues (calm, stressed, excited) and then reflects that back as a 'mood flower' plus a one-sentence insight.
• two inputs: voice + text
• for voice: record locally with AVAudioSession, transcribe with Whisper (API to start, the. consider on-device) and run lightweight sentiment/emotion classifier (either small CoreML classifier on the transcriopt or simple API sentiment while I prototype)
• Mapping scores to visuals and store entries locally with Core Data.
• V1 features include: record, analyze, mood flower + note, history timeline with weekly trends

1. ClearVoice - Real-time Speech Aid

• supportive practice tool for clearer, calmer speech
• useful for interviews, presentations, or language learning
• gives actionable feedback and a friendly readasbility suggestion for a sentence I'm about to say
• while I speak, I stream audio frames and either do a low-latency on-device metrics
• run lightweight detectors

Progress

Currently, MindMeld is in its draft and prototype phase. Most key features have been implemented, including voice recording, text input, sentiment analysis, mood visualization, local storage, and basic user interface components. However, several technical issues remain to be addressed before the project will be production-ready. These challenges include the configuration of the Core Data model, such as manually creating entities and duplicate files; build errors raised because of file conflicts in Xcode; basic error handling with limited offline support; and secure API key management, where API keys are not stored securely. Resolving these issues and refining this implementation, especially the build and Core Data integration, are the top priorities for the next development steps.

The roadmap for the project has several enhancements envisioned for the phases of work to come. Immediately, goals include stabilizing the build, improving data model recognition, testing across iOS devices, and general code clean-up. Short-term improvements include on-device transcription and a CoreML-based emotion classifier for improved privacy and offline processing, advanced mood flower visualizations, and increased data export options. Medium-term plans include upgrading the user experience with reminders, deeper analytics, and personalization features, while longer-term ambitions revolve around security, optimizing performance, overall testing, and preparation for App Store release. Longer-term, MindMeld will evolve into a fully privacy-focused, on-device emotional wellness companion that meets Apple's standards for privacy and local data processing.