Poem meter classification of spoken Arabic poetry: Integrating high-resource systems for a low-resource task
Abstract
Arabic poetry constitutes an essential and integral component of the Arabic language and cultural heritage. Historically, Arabs have used poetry to document significant events, including battles and conflicts, as well as to express romance, pride, lamentation, and various other sentiments. Over the decades, Arabic poetry has attracted substantial scholarly attention from linguists. A distinguishing feature of Arabic poetry is its distinctive rhythmic structure, known as meter, which differentiates it from prose. These meters, along with other poetic characteristics, are extensively studied in the Arabic linguistic field of "Aroud." Identifying the meter of a verse is a complex and time-consuming process requiring specialized knowledge in Aroud. For spoken poetry, this process becomes even more challenging due to the additional layer of audio processing. In this study, we propose a state-of-the-art framework to identify poem meters in recited Arabic poetry by integrating two separate high-resource systems to perform this low-resource task. To facilitate future research and ensure generalization of our proposed architecture, we publish a benchmark dataset specifically designed for this task.