Casablanca: Data and Models for Multidialectal Arabic Speech Recognition

TL;DR

This paper introduces Casablanca, a large-scale, community-driven dataset for eight Arabic dialects, addressing the lack of diverse speech data and enabling improved speech recognition models for underrepresented dialects.

Contribution

The creation of Casablanca, a comprehensive multi-dialectal Arabic speech dataset with annotations, and the development of baseline models leveraging this resource.

Findings

Casablanca covers eight Arabic dialects with extensive annotations.

Baseline models demonstrate the dataset's utility for speech recognition.

The dataset facilitates future research in dialectal Arabic speech processing.

Abstract

In spite of the recent progress in speech processing, the majority of world languages and dialects remain uncovered. This situation only furthers an already wide technological divide, thereby hindering technological and socioeconomic inclusion. This challenge is largely due to the absence of datasets that can empower diverse speech systems. In this paper, we seek to mitigate this obstacle for a number of Arabic dialects by presenting Casablanca, a large-scale community-driven effort to collect and transcribe a multi-dialectal Arabic dataset. The dataset covers eight dialects: Algerian, Egyptian, Emirati, Jordanian, Mauritanian, Moroccan, Palestinian, and Yemeni, and includes annotations for transcription, gender, dialect, and code-switching. We also develop a number of strong baselines exploiting Casablanca. The project page for Casablanca is accessible at: www.dlnlp.ai/speech/casablanca.