Skip to content

Dataset · Muno459

Arabic speech dataset for ASR

Muno459/QuranLab-Arabic-Speech

Wideband Arabic speech in segments for training ASR and TTS models.

What's inside

Contains segments of broad-spectrum Arabic speech ranging from 2 to 20 seconds in 48 kHz Opus format. Audio is segmented by natural pauses, silences of 1.5s or longer are removed, and access is restricted. Size ranges from 10M to 100M items.

What you can do with it

  • Training and fine-tuning automatic speech recognition (ASR) systems
  • Developing text-to-speech (TTS) models
  • Transcription using LLMs
Tasks
Automatic speech recognition, Text → speech
Languages
Arabic
Size
10M–100M rows
Downloads this month
21.8K
Likes
1
Access
on request on Hugging Face

How to load it

from datasets import load_dataset

ds = load_dataset("Muno459/QuranLab-Arabic-Speech")