Cloud-Native Voice Translator
Transitioning from ambiguous user needs to a deployed cloud solution. A Product Owner perspective on defining system boundaries, managing Azure Cognitive Services quotas, and executing a precise Waterfall lifecycle.
My Role
Full Stack Dev & Project Lead
Authored the SRS v1.0, managed the Gantt timeline, and implemented the Xamarin Android client.
The Constraint
Azure Free Tier Limits:
Strict quota of 5 hours audio/month. The architecture had to handle "Quota Exceeded" exceptions gracefully without crashing.
The Outcome
10-week Waterfall cycle completed with full documentation.
Logic Flow & Architecture
Based on the Software Requirements Specification (SRS), I designed a synchronous flow that toggles between Voice and Text inputs.
SRS-REQ-01: Latency
"If Azure API fails to respond > 3000ms, the local 'Result' module must trigger the Error State workflow."
SRS-INT-02: Data Privacy
"Application shall not store personal voice data persistently. Audio buffers must be cleared from memory immediately after result rendering."
Waterfall Execution
Managed a 10-week development cycle. The critical path relied on finalizing the SRS before the "Environment Setup" phase to avoid costly cloud reconfiguration later.
Mobile Implementation
Clean, Single-Page Interface
Adhering to the SRS requirement for a "one-screen application," the UI is split vertically: Input (Top) and Output (Bottom).
-
Bilingual Support
Dynamic toggling between source and target languages.
-
Visual Feedback
Microphone icon pulses during active listening state.
English
Hello, nice to meet you.French
Bonjour, ravi de vous rencontrer.