Gherkin feature files defining test scenarios

Feature: Onboarding and Consent As a new user I want to go through onboarding, language selection, and accept the privacy notice So that I can use the app with full consent and my preferred language Scenario: User accepts onboarding and consent Given the app is launched for the first time When I view the onboarding carousel And I accept the EULA and privacy notice Then I should proceed to the language selection screen Scenario: User rejects onboarding and consent Given the app is launched for the first time When I decline the EULA and privacy notice Then the app should exit Scenario: User changes the language Given I am on the language selection screen When I select "Spanish" Then the app should display all subsequent screens in Spanish Feature: Model Selection and Initialization As a new user I want to choose and download a compatible ECAPA-TDNN model So that the app is ready for offline analysis Scenario: User selects and downloads model Given I am on the model selection screen When I choose the Small model and download it Then the model is saved and initialized for inference Scenario: Model integrity check on start Given a model is already downloaded When the app starts Then the model checksum is verified before first use Feature: Home Dashboard and Navigation As a user I want to see my current stress level and quickly access core features So that I can navigate the app efficiently Scenario: Home dashboard shows stress status and actions Given I have completed onboarding and model setup When I arrive at the home screen Then I should see my latest stress result and a voice check button And there are quick links to History and Settings Feature: Voice Stress Check As a user I want to record my voice and get a stress result So that I can monitor my wellbeing in real-time Scenario: User records and submits a voice sample Given I am on the home screen When I tap the "Voice Check" button And I record up to 30 seconds of audio And I stop the recording Then the app should process my recording and display a stress result and confidence Scenario: User cancels recording Given I am in the voice recording overlay When I tap "Cancel" Then the recording stops and I return to the home screen Feature: Result Display and Recommendations As a user I want to see my stress score and relevant advice after each assessment So that I can take appropriate action Scenario: Stress result and advice are displayed Given I have just completed a voice assessment Then my stress score (1-10), confidence, and a suggestion are shown And I can tag the context (e.g., "Meeting") And I can save the result Feature: History and Trend Visualization As a user I want to view a list and simple chart of recent stress assessments So that I can see my mental health trends over time Scenario: Viewing the history list Given I am on the home screen When I tap the History button Then I see a list of my past stress scores with dates and context Scenario: Viewing the trend chart Given I am on the history screen When I view the screen Then I see a line chart of my last 7 days of stress scores Feature: Settings and Privacy As a user I want to configure basic preferences and manage my data So that I have control over appearance, models, and privacy Scenario: Changing app language or theme Given I am in the settings screen When I change the language to "French" or toggle the theme Then the change applies immediately to all screens Scenario: Managing models Given I am in the settings screen When I view model information Then I see the version, storage used, and option to re-download Scenario: Exporting and deleting data Given I am in the settings screen When I export my data Then I receive a portable archive of my assessments When I delete my data Then all stored assessments should be erased Feature: Performance and Hardware As a system administrator I want the app to detect the correct delegate at startup So that inference is optimal on all supported devices Scenario: NPU preference is selected if available Given the device supports NNAPI or Qualcomm QNN When the app starts Then model inference uses the NPU delegate Scenario: Fallback to GPU or CPU as required Given no NPU is available When the app starts Then model inference uses the GPU or CPU delegate Scenario: Inference under 3 seconds Given I record a valid audio sample When processing occurs Then the result is available within 3 seconds Feature: Offline Support and Data Security As a user I want my data to remain on-device and for all features to work offline So that my privacy is protected and the app is reliable Scenario: App works offline after initial setup Given a model is already downloaded When I use any main feature with no network Then all processing and storage remains fully functional Scenario: Data export produces encrypted user archive Given I request a data export in Settings Then my data is packaged as an encrypted, portable file