Seif-Eldeen-Sameh/asr_codeswitched_dataset
Arabic/English Code-Switched ASR Dataset Audio + transcripts of code-switched Egyptian Arabic and English speech, assembled to fine-tune ASR for Arab-world lecture content where dialectal Arabic and English technical vocabulary alternate within sentences. Composition Source Description EJUST custom recordings Locally recorded/segmented code-switched clips (segments_codeswitched.csv + WAVs) MohamedRashad/arabic-english-code-switching ~12,480 public… See the full description on the dataset page: https://huggingface.co/datasets/Seif-Eldeen-Sameh/asr_codeswitched_dataset.
<!-- Upload this as the README.md of https://huggingface.co/datasets/Seif-Eldeen-Sameh/asrcodeswitcheddataset -->
Arabic/English Code-Switched ASR Dataset
Audio + transcripts of code-switched Egyptian Arabic and English speech, assembled to fine-tune ASR for Arab-world lecture content where dialectal Arabic and English technical vocabulary alternate within sentences.
Composition
Total: 34.1 hours · 45,189 clips · avg 2.7 s/clip (utterance-level, ideal for Whisper-class ASR).
Schema
Load
from datasets import load_dataset, Audio
ds = load_dataset("Seif-Eldeen-Sameh/asr_codeswitched_dataset", split="train")
ds = ds.cast_column("audio", Audio(sampling_rate=16000))Intended use
Fine-tuning ASR (e.g. Whisper) for code-switched Arabic/English. See the model trained on it: `whisper-medium-arabic-codeswitched`.
Limitations & ethics
- Egyptian dialect dominant; not representative of all Arabic dialects.
- Lecture/talk domain; technical vocabulary skew.
- Ensure you have rights to any audio you add; respect speaker consent.
License
MIT (custom portion). The public source dataset retains its own license — review before redistribution.