Android AI Screen-Reader Assistant
Budget: ₹1,500 – ₹12,500 INR
I’m building a Jarvis-style companion for Android that can look at whatever is on-screen, understand it, and talk back to the user. The core flow is simple: an accessibility service captures the visible text, sends it to GPT for comprehension, and returns a concise spoken summary. At any moment the user can also issue a voice command (“scroll down”, “explain this graph”, “copy code block”) and the app should act or respond accordingly.
Key points I need you to cover
• Minimalistic interface: a small floating button or edge swipe that invokes the assistant without cluttering the screen.
• Voice interaction end-to-end: speech-to-text for commands, text-to-speech for replies.
• Screen reading & summarisation: pull structured content from any foreground app, pipe it to GPT (OpenAI API) and deliver a short, clear summary.
• Google Assistant hand-off: if the user says “open maps to home”, route the intent to Google Assistant rather than handling it locally.
• Privacy-first: request only the permissions absolutely required by the Accessibility API and clearly signal when capture is active.
Deliverables
1. Production-ready Android Studio project (Kotlin preferred) with clean, well-commented code.
2. README explaining setup of OpenAI keys, Google actions, and any Firebase/Crashlytics you add.
3. Signed APK for quick testing plus instructions for Play Store deployment.
4. One-minute demo video showing voice command, on-screen summarisation, and Assistant hand-off in action.
I’ll consider the job complete when the APK runs smoothly on Android 10–14, performs the above flows with <2 s latency on a mid-range device, and passes a quick code review for clarity and modularity.
Key points I need you to cover
• Minimalistic interface: a small floating button or edge swipe that invokes the assistant without cluttering the screen.
• Voice interaction end-to-end: speech-to-text for commands, text-to-speech for replies.
• Screen reading & summarisation: pull structured content from any foreground app, pipe it to GPT (OpenAI API) and deliver a short, clear summary.
• Google Assistant hand-off: if the user says “open maps to home”, route the intent to Google Assistant rather than handling it locally.
• Privacy-first: request only the permissions absolutely required by the Accessibility API and clearly signal when capture is active.
Deliverables
1. Production-ready Android Studio project (Kotlin preferred) with clean, well-commented code.
2. README explaining setup of OpenAI keys, Google actions, and any Firebase/Crashlytics you add.
3. Signed APK for quick testing plus instructions for Play Store deployment.
4. One-minute demo video showing voice command, on-screen summarisation, and Assistant hand-off in action.
I’ll consider the job complete when the APK runs smoothly on Android 10–14, performs the above flows with <2 s latency on a mid-range device, and passes a quick code review for clarity and modularity.