In the fast-paced world of business, numerous phone calls and meetings occur where critical decisions are made by various participants. Traditionally, someone would manually take notes during these discussions and distribute them to attendees to confirm the outcomes. By leveraging voice-to-text technologies and implementing AI capable of understanding the conversation’s context, it is possible to automate the meeting summary process efficiently. 

Features:

1. Voice-Text Conversion

2. Multiple Speaker Detection

3. Multiple Language Detection and Translation

4. Extractive Text Summarisation Ensembling using NLP PageRank and Sentence Embeddings.

5. Meeting/ Phone Analytics.

6. Minutes of the meeting are sent using EMAIL to the participants.

Technologies:

1. Web APP - HTML, CSS, JS, Flask, Python

2. Voice -> Text/Multiple Speaker Detection - IBM Watson Speech - text -API, REST

3. Multiple Language Detection and Translation - Google Translate

4. Extractive Text Summarisation Ensembling using NLP PageRank and Sentence Embeddings - Numpy, NLTK

5. Meeting Analytics - Pandas, d3, Gensim, word cloud, spacy, Topic Modelling using LDA.

Thousands of ICOs(initial coin offerings are a form of cryptocurrency that businesses use to raise capital) are launched every year with white papers outlining their vision and strategy. A massive amount of these projects are scams designed to steal money from investors. Distinguishing between credible projects from scams or poorly conceived ideas is critical for investors and analysts. This project leverages state-of-the-art natural language processing and deep learning-based AI models to efficiently automate the classification of ICO white papers. The classifier identifies patterns to determine a project's legitimacy or focus area by analyzing context, technical terminology, sentiment, and readability. This automated approach streamlines the evaluation process, ensuring stakeholders can make informed decisions swiftly and accurately. 

Key Features:

1. Web Scraping and White Paper Pre-Processing: Scrape Icobench.com for new cryptocurrencies and get their whitepaper.

2. Contextual Classification: Distinguishes between legitimate, fraudulent, and technology-driven ICOs.

3. Sentiment Analysis: Detects overly optimistic or suspicious language patterns.

4. Keyword Extraction: Identifies blockchain-specific terminology and critical project details.

5. Automated Evaluation: Reduces the need for manual analysis, speeding up decision-making.

Technologies and Tools:

1. Web Scraping: Beautiful Soup, and Python for web scraping.

2. NLP Models: Deep-learning-based models for classifications and contextual understanding.

3. Data Processing: Numpy, Pandas, NLTK and spaCy for feature extraction and text preprocessing.

4. Classifier: Assigns a risk percentage for the whitepapers for better analysis and investment decisions.

The Simpsons is a cultural phenomenon known for its unique blend of humor, character dynamics, and pop culture references. Creating an automated script generator using RNN neural networks offers a way to explore the intersection of creativity and AI, enabling fans and writers to generate their own Simpsons-style episode snippets with the characters from the bar in Simpsons universe. 

Features:

• Script Generation: Generates humorous and coherent Simpsons-style episode scripts.

• Character-Driven Dialogue: Captures the unique speech styles of characters like Homer, Moe, and Barney.

• Dynamic Themes: Users can input custom themes and characters to personalize the generated content.

• Sequential Text Modeling: Uses RNN to predict the next word in the sequence based on prior context.

• Customizable Text Length: Allows users to control the snippet length for tailored outputs.

Technologies:

• Model Training and Text Generation: Python, TensorFlow

• Text Preprocessing: NLTK, NumPy, Tokenizer.

• Data Analytics and Evaluation: Pandas for data handling and evaluation metrics