JTC-Hackathon-Team-1/Beta1

first beta implementation of the code to run

0

stars

0

commits

Python

primary language

Apr 24, 2025

updated

README

CasaLingua v0.21.0

🧠 Overview

CasaLingua is a multilingual AI voice and text assistant for simplifying housing applications. It automatically processes user inputs—whether audio, image, or text—and outputs readable, simplified forms with ethical compliance and audit capability.


📦 Modules Overview

1. app/input_funnel.py

  • Detects file types: .txt, .pdf, .jpg/.png, .wav/.mp3
  • Routes input to OCR or ASR processors

2. app/text_pipeline.py

  • Orchestrates NER, translation, and governance checks

3. app/translation_llm.py

  • Translates text to English or another target language
  • Simulates simplification for legal or complex phrases

4. app/ner_module.py

  • Extracts entities like name, address, DOB for form auto-fill

5. app/governance_llm.py

  • Applies ethical audits (bias, fidelity, PII compliance)
  • Returns a confidence score and alignment check

6. app/utils/ocr.py

  • Extracts text from image or PDF using placeholder OCR

7. app/utils/audio_tools.py

  • Transcribes audio to text using placeholder ASR

🧪 Tests

  • tests/test_pipeline.py: Validates end-to-end output from raw text input

🚀 Running the App

python main.py

🎛️ Admin Panel (Optional)

uvicorn admin_panel.main:app --reload

🔍 Future Features

  • Whisper integration
  • PDF input support with OCR confidence
  • Multilingual conversation agent with memory

🧠 Logic Ladder Diagram

graph TD
    A[User Input (Audio/Text/Image)] --> B[Input Funnel]
    B --> C[Text Pipeline]
    C --> D[NER Extraction]
    C --> E[Translation Module]
    C --> F[Governance & Audit]
    D & E & F --> G[Final Output & Audit Report]

JTC-Hackathon-Team-1/Beta1

first beta implementation of the code to run

0

stars

0

commits

Python

primary language

Apr 24, 2025

updated

README

CasaLingua v0.21.0

🧠 Overview

CasaLingua is a multilingual AI voice and text assistant for simplifying housing applications. It automatically processes user inputs—whether audio, image, or text—and outputs readable, simplified forms with ethical compliance and audit capability.


📦 Modules Overview

1. app/input_funnel.py

  • Detects file types: .txt, .pdf, .jpg/.png, .wav/.mp3
  • Routes input to OCR or ASR processors

2. app/text_pipeline.py

  • Orchestrates NER, translation, and governance checks

3. app/translation_llm.py

  • Translates text to English or another target language
  • Simulates simplification for legal or complex phrases

4. app/ner_module.py

  • Extracts entities like name, address, DOB for form auto-fill

5. app/governance_llm.py

  • Applies ethical audits (bias, fidelity, PII compliance)
  • Returns a confidence score and alignment check

6. app/utils/ocr.py

  • Extracts text from image or PDF using placeholder OCR

7. app/utils/audio_tools.py

  • Transcribes audio to text using placeholder ASR

🧪 Tests

  • tests/test_pipeline.py: Validates end-to-end output from raw text input

🚀 Running the App

python main.py

🎛️ Admin Panel (Optional)

uvicorn admin_panel.main:app --reload

🔍 Future Features

  • Whisper integration
  • PDF input support with OCR confidence
  • Multilingual conversation agent with memory

🧠 Logic Ladder Diagram

graph TD
    A[User Input (Audio/Text/Image)] --> B[Input Funnel]
    B --> C[Text Pipeline]
    C --> D[NER Extraction]
    C --> E[Translation Module]
    C --> F[Governance & Audit]
    D & E & F --> G[Final Output & Audit Report]

Languages

Python

99.5%