Habibi: Laying the Open-Source Foundation of Unified-Dialectal Arabic Speech Synthesis

TL;DR

Habibi introduces a unified open-source Arabic TTS system covering 12+ dialects, utilizing a novel curriculum learning approach and a new benchmark, achieving high-quality synthesis comparable to commercial models.

Contribution

It presents the first comprehensive multi-dialect Arabic TTS framework with a standardized benchmark and open-source resources, addressing key challenges in dialectal diversity and data scarcity.

Findings

Unified model matches or surpasses specialized dialect models.

Achieves high intelligibility, speaker similarity, and naturalness.

Validated through extensive ablations and human evaluations.

Abstract

Arabic spans over 30 spoken varieties, yet no open-source text-to-speech system unifies them. Key barriers include substantial cross-dialect lexical and phonological divergence, scarce synthesis-grade data, and the absence of a standardized multi-dialect evaluation benchmark. We present Habibi, a unified-dialectal Arabic TTS framework that addresses all three. Through a multi-step curation pipeline, we repurpose open-source ASR corpora into TTS training data covering 12+ regional dialects. A linguistically-informed curriculum learning strategy - progressing from Modern Standard Arabic to dialectal data - enables robust zero-shot synthesis without text diacritization. We further release the first standardized multi-dialect Arabic TTS benchmark, comprising over 11,000 utterances across 7 dialect subsets with manually verified transcripts. On this benchmark, our unified model matches or surpasses per-dialect specialized models. Both automatic metrics and human evaluations confirm that Habibi is highly competitive with ElevenLabs' Eleven v3 (alpha) in intelligibility, speaker similarity, and naturalness. Extensive ablations (~8,000 H100 GPU hours, 30+ configurations) validate each design choice. We open-source all checkpoints, training and inference code, and benchmark data - the first such release for multi-dialect Arabic TTS - at https://SWivid.github.io/Habibi/ .