Elyadata/AraBEST-RQ-300M-6k

Ara-BEST-RQ-300M-6k

Ara-BEST-RQ-300M-6k is a 300M-parameter self-supervised speech representation model for Arabic and Arabic dialects. It is part of the Ara-BEST-RQ family introduced in Ara-Best-RQ: Multi Dialectal Arabic SSL.

This model was pretrained on the crawled Ara-BEST-RQ dataset: approximately 5.6k hours of Creative Commons Arabic speech collected from publicly available YouTube videos and segmented for self-supervised speech learning.

• Paper: Ara-Best-RQ: Multi Dialectal Arabic SSL
• Dataset: Elyadata/Ara-Best-RQ_dataset
• Implementation: elyadata/AraBEST-RQ

Model Details

Model Description

Ara-BEST-RQ is a family of Arabic-focused self-supervised learning (SSL) speech models based on the BEST-RQ framework. The models are designed to learn speech representations that transfer well to Arabic speech processing tasks, including automatic speech recognition (ASR) and dialect identification (DID).

This checkpoint corresponds to the 300M variant pretrained on the crawled 6k-hour dataset.

• Model type: Self-supervised speech representation model
• Architecture: Conformer-based BEST-RQ encoder
• Parameters: ~300M
• Training data: Crawled Arabic speech data
• Training hours: ~5,640 hours
• Languages: Arabic, including multiple dialects
• Primary use: Speech representation learning / downstream fine-tuning

Architecture

The 300M Ara-BEST-RQ model uses:

• 24 Conformer encoder layers
• Model dimension: 848
• 8 attention heads
• Feed-forward dimension: 2048
• GELU activations
• Relative position multi-head attention
• Convolutional front-end
• Random projection quantizer with 4096 codebook entries

Training Data

The model was pretrained on the crawled Ara-BEST-RQ dataset.

The released dataset on Hugging Face provides metadata only: YouTube video identifiers and audio segment boundaries. No audio or video files are distributed as part of the dataset.

Dataset link: Elyadata/Ara-Best-RQ_dataset

Evaluation

The paper evaluates Ara-BEST-RQ models on automatic speech recognition and dialect identification tasks. The following results are reported for the Ara-BEST-RQ crawled 300M model.

Automatic Speech Recognition

WER scores on ASR benchmarks:

Dataset	WER
Common Voice 19.0 Arabic	18.67
MGB-3	30.85
MGB-5	54.18
TARIC-SLU	23.98
Average	31.92

Dialect Identification

Results on ADI-20:

Split	Accuracy	Weighted F1
Validation	97.21	97.17
Test	96.02	95.98

Usage

This is a self-supervised pretrained model intended to be used as a speech encoder or as an initialization checkpoint for downstream fine-tuning.

For training and fine-tuning recipes, please refer to the official implementation:

git clone https://github.com/elyadata/AraBEST-RQ
cd AraBEST-RQ

You can download the checkpoint from Hugging Face using:

from huggingface_hub import snapshot_download

model_dir = snapshot_download("Elyadata/AraBEST-RQ-300M-6k")
print(model_dir)

Please refer to the repository configuration and SpeechBrain recipes for the correct model-loading interface.

Fine-tuning with SpeechBrain

To fine-tune this pretrained Ara-BEST-RQ checkpoint in a SpeechBrain recipe, adapt the pretrainer section of your YAML configuration so that it loads both the pretrained model checkpoint and the corresponding normalizer.

Example:

pretrainer: !new:speechbrain.utils.parameter_transfer.Pretrainer
    collect_in: !ref <save_folder>
    loadables:
        pt_model: !ref <pt_model>
        normalize: !ref <normalize>
    paths:
        pt_model: !ref <pt_model_path>/model.ckpt
        normalize: !ref <pt_model_path>/normalizer.ckpt

In your downstream recipe, make sure that:

• <pt_model> points to the Ara-BEST-RQ pretrained model object used in your training graph.
• <normalize> points to the normalization module used by the recipe.
• <pt_model_path> points to the local directory containing model.ckpt and normalizer.ckpt.
• <save_folder> is the experiment directory where SpeechBrain should collect and manage pretrained components.

This setup allows SpeechBrain to initialize the downstream model from the Ara-BEST-RQ SSL checkpoint before fine-tuning on task-specific data.

Citation

If you use this model, please cite the Ara-BEST-RQ paper:

@misc{elleuch2026arabestrqmultidialectalarabic,
      title={Ara-Best-RQ: Multi Dialectal Arabic SSL}, 
      author={Haroun Elleuch and Ryan Whetten and Salima Mdhaffar and Yannick Estève and Fethi Bougares},
      year={2026},
      eprint={2603.21900},
      archivePrefix={arXiv},
      primaryClass={cs.CL},
      url={https://arxiv.org/abs/2603.21900}, 
}