---
dataset_info:
  features:
  - name: audio
    dtype: audio
  - name: text
    dtype: string
  - name: cleaned_text
    dtype: string
  - name: environment_type
    dtype: string
  - name: speaker_id
    dtype: string
  - name: speaker_gender
    dtype: string
  splits:
  - name: test
    num_bytes: 2702664819.072
    num_examples: 3296
  download_size: 2564183776
  dataset_size: 2702664819.072
configs:
- config_name: default
  data_files:
  - split: test
    path: data/test-*
task_categories:
- automatic-speech-recognition
- text-to-speech
language:
- ar
pretty_name: SCC 2022
size_categories:
- 1K<n<10K
license: cc-by-nc-sa-4.0
---

<center>
  <img src="https://cdn-uploads.huggingface.co/production/uploads/6116d0584ef9fdfbf45dc4d9/5g0ERAppFLxQhXU2Haj_X.png"/>
</center>

# Dataset Card for Saudilang Code-Switch Corpus (SCC)

## Dataset Summary

The **Saudilang Code-Switch Corpus (SCC)** is a **5-hour** transcribed audio dataset featuring informal Saudi Arabic speech with **code-switching to English**, sourced from the **"Thmanyah" YouTube podcast**. It was created for evaluating models in multilingual and dialectal speech settings, such as **ASR**, **language identification**, and **code-switch detection**.

## Supported Tasks and Leaderboards

This dataset is suitable for testing models in:

* **Automatic Speech Recognition (ASR)**
* **Text-to-Speech (TTS)**
* **Speaker Diarization**
* **Language Identification**
* **Dialect Detection**
* **Code-Switching Detection**

## Languages

* **Arabic (ar)** – primarily Saudi dialect
* **English (en)** – code-switched within Arabic speech

## Dataset Structure

### Data Fields

* `audio`: The raw audio recording (e.g., `.wav`)
* `text`: The original transcription including both Arabic and English content
* `cleaned_text`: A normalized version of the transcription
* `environment_type`: Acoustic environment in which the speech was recorded (Clean, Music, or Noisy)
* `speaker_id`: An anonymized identifier for the speaker
* `speaker_gender`: Gender of the speaker (all of them are Male)

### Splits

* **Test Set Only**: \~5 hours of annotated, transcribed speech
  *(This dataset is designed primarily for evaluation purposes.)*

## Dataset Creation

### Source and Curation

The dataset was curated from naturally occurring podcast conversations and **transcribed by the National Center for Artificial Intelligence at SDAIA**.

### Motivation

Given the scarcity of high-quality Arabic-English code-switching resources, this dataset was developed to address that gap and enable progress in speech technologies for bilingual Arabic users. Publishing SCC demonstrates a commitment to enriching Arabic digital resources and enabling the development of AI models that are more linguistically inclusive and reflective of real-world usage.

## Licensing

This dataset is licensed under the **Creative Commons Attribution-NonCommercial-ShareAlike 4.0 International (CC BY-NC-SA 4.0)** license.
More details: [https://creativecommons.org/licenses/by-nc-sa/4.0/](https://creativecommons.org/licenses/by-nc-sa/4.0/)

## Citation

```
@misc{SCC2025,
  title={Saudilang Code-Switch Corpus (SCC)},
  author={SDAIA},
  year={2022},
  howpublished={\url{https://www.kaggle.com/datasets/sdaiancai/saudilang-code-switch-corpus-scc}},
  note={CC BY-NC-SA 4.0}
}
```

## Contributions

Dataset curated by **SDAIA**. Dataset card written by the open-source community.