首个统一阿拉伯多方言语音合成框架，解决数据少与评测难问题。
Habibi: Laying the Open-Source Foundation of Unified-Dialectal Arabic Speech Synthesis
- 用多步骤清洗流程将开源语音识别数据转为多方言语音训练集。
- 零样本合成效果超越单方言模型，支持无分词文本直接生成。
- 开源完整代码、模型和首个多方言评测基准，适合研究者使用。
阿拉伯语有超过30种口语变体，但尚无开源的统一语音合成系统。主要障碍包括方言间词汇与发音差异大、高质量合成数据稀缺，以及缺乏标准化的多方言评估基准。我们提出Habibi，一个统一阿拉伯多方言语音合成框架，解决了上述三大挑战。通过多阶段数据清洗流程，我们将开源语音识别语料库转化为覆盖12种以上地区方言的语音合成训练数据。采用语言学启发的渐进式课程学习策略——从现代标准阿拉伯语逐步过渡到方言数据——使模型无需文本分词即可实现鲁棒的零样本合成。我们还发布了首个标准化的多方言阿拉伯语语音合成评测基准，包含7个方言子集，共超过11,000条经人工验证的语音片段。在该基准上，统一模型表现达到或优于各方言专用模型。自动指标与人工评测均表明，Habibi在可懂度、说话人相似性和自然度方面与ElevenLabs Eleven v3（alpha）相当。大规模消融实验（约8,000小时H100 GPU计算，30+配置）验证了每个设计选择的有效性。我们已将所有检查点、训练与推理代码及基准数据开源，地址为https://SWivid.github.io/Habibi/，这是首个针对多方言阿拉伯语语音合成的完整开源发布。
原文摘要 · Abstract (English)
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/ .
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。