# End-of-Utterance (EOU) Dataset
## Overview
This dataset is designed for End-of-Utterance (EOU) detection in conversational speech and text.
The goal is to determine whether a given text segment represents the end of a speaker’s turn.
The dataset is suitable for training models used in real-time dialogue systems such as voice assistants and streaming conversational agents.
---
## Task Definition
Given a text segment, predict a binary label:
- EOU = 1 → The speaker has finished their turn
- EOU = 0 → The speaker is expected to continue speaking
---
## Dataset Format
The dataset is provided in CSV format.
Each row corresponds to a single text segment.
### Columns
- text: Input text segment
- eou: Binary label (1 = end of utterance, 0 = not end)
---
## Language
- Primary language: Arabic (Saudi)
- The dataset includes:
- Partial utterances
- Complete sentences
- Conversational fragments
- Natural spoken-style text
---
## Intended Use
- End-of-turn detection
…
