OTS-Dataset System Requirements Document

Introduction

The OTS-Dataset project aims to develop a website for selling the Speech OTS (Off The Shelf) dataset. The primary customers are companies building AI models for speech recognition and processing. The website will offer data previews and a single licensing model for purchasing the dataset.

System Overview

The OTS-Dataset website will serve as a platform for companies to purchase speech datasets. It will provide a user-friendly interface for browsing, previewing, and licensing datasets. The system will ensure secure transactions and manage customer accounts efficiently.

Functional Requirements

• As a Company, I should be able to browse available speech datasets.
• As a Company, I should be able to preview a sample of the dataset before purchasing.
• As a Company, I should be able to purchase a dataset through a single licensing model.
• As a Company, I should be able to view licensing options clearly.
• As a Company, I should be able to securely complete transactions on the website.
• As a Company, I should be able to create and manage my account on the website.

User Personas

• Company Representative: A user representing a company interested in purchasing speech datasets for AI model development. They require access to dataset previews and licensing information.

Core User Flows

• Company Representative browses datasets -> previews dataset sample -> selects licensing option -> completes purchase -> downloads dataset.
• Company Representative creates account -> logs in -> manages account details -> views purchase history.

Visuals Colors and Theme

• primary: #1A73E8 (deep blue)
• primary_light: #E8F0FE (light blue)
• secondary: #FF6F61 (coral)
• accent: #FFD700 (gold)
• highlight: #FFA500 (orange)
• bg: #FFFFFF (white)
• surface: rgba(240, 240, 240, 0.8)
• text: #202124 (dark gray)
• text_muted: #5F6368 (light gray)
• border: rgba(200, 200, 200, 0.5)

Signature Design Concept

The homepage will feature an interactive 3D globe using @react-three/fiber and @react-three/drei. Users can rotate the globe to explore different regions where speech datasets are available. Clicking on a region will zoom in to reveal dataset details and previews. The globe will have dynamic lighting effects that change based on the time of day, providing a visually engaging experience.

Interaction Model & Motion Direction

• Interaction Model: Parallax
  • The homepage will use a parallax scrolling effect to create depth, with layered content moving at different speeds. This will enhance the storytelling aspect of the dataset offerings.
• Internal Pages: Static
  • Internal pages such as account management and transaction history will focus on clarity and minimal motion to ensure ease of use.

Non-Functional Requirements

• The website must ensure secure data transactions and protect user information.
• The system should be scalable to handle increasing numbers of datasets and users.
• The website should be responsive and accessible across various devices and screen sizes.

Tech Stack

• Frontend: React for Web
• Backend: Python, FastAPI
• Database: MySQL or MariaDB, using Alembic for migrations
• AI Models: Not applicable
• AI Tools: Not applicable
• Local Orchestration: Docker, docker-compose
• Server-side Orchestration: Kubernetes

Assumptions and Constraints

• The website will initially support a single licensing model.
• The primary user base will be companies involved in AI model development.
• The system will comply with local data protection regulations in India.

Glossary

• OTS (Off The Shelf): Refers to datasets that are ready for immediate use without customization.
• Licensing Model: A framework that defines how datasets can be purchased and used by customers.
• Parallax: A visual effect where background images move slower than foreground images, creating a sense of depth.