mlsanityData to 14 Oct 2026

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

arXiv 2601.13802 ↗First posted 20 Jan 2026cs.CL cs.SD eess.AScited by 3

Not on the map. Why?

Abstract (arXiv)

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/ .

Key facts. Cited by 3 papers in the map (data through 14 Oct 2026). Builds on 33 of its 45 references that are in the map. First posted on arXiv 20 Jan 2026.

Builds on

33 of 45 references.

References · 33

  1. 2023Neural Codec Language Models are Zero-Shot Text to Speech Synthesizerscited by 752
  2. 2022FLEURS: Few-shot Learning Evaluation of Universal Representations of Speechcited by 469
  3. 2023Voicebox: Text-Guided Multilingual Universal Speech Generation at Scalecited by 379
  4. 2024F5-TTS: A Fairytaler that Fakes Fluent and Faithful Speech with Flow Matchingcited by 361
  5. 2024CosyVoice: A Scalable Multilingual Zero-shot Text-to-speech Synthesizer based on Supervised Semantic Tokenscited by 334
  6. 2024Seed-TTS: A Family of High-Quality Versatile Speech Generation Modelscited by 300
  7. 2024MaskGCT: Zero-Shot Text-to-Speech with Masked Generative Codec Transformercited by 161
  8. 2024E2 TTS: Embarrassingly Easy Fully Non-Autoregressive Zero-Shot TTScited by 152
  9. 2025Spark-TTS: An Efficient LLM-Based Text-to-Speech Model with Single-Stream Decoupled Speech Tokenscited by 113
  10. 2021Large-scale Self-Supervised Speech Representation Learning for Automatic Speaker Verificationcited by 78
  11. 2022Music Source Separation with Band-split RNNcited by 70
  12. 2024FireRedTTS: A Foundation Text-To-Speech Framework for Industry-Level Generative Speech Applicationscited by 68
21 more
  1. 2024Autoregressive Speech Synthesis without Vector Quantizationcited by 65
  2. 2017Speech Recognition Challenge in the Wild: Arabic MGB-3cited by 60
  3. 2025DiTAR: Diffusion Transformer Autoregressive Modeling for Speech Generationcited by 57
  4. 2025MiniMax-Speech: Intrinsic Zero-Shot Text-to-Speech with a Learnable Speaker Encodercited by 48
  5. 2021QASR: QCRI Aljazeera Speech Resource -- A Large Scale Annotated Arabic Speech Corpuscited by 38
  6. 2023E3 TTS: Easy End-to-End Diffusion-based Text to Speechcited by 30
  7. 2025ZipVoice: Fast and High-Quality Zero-Shot Text-to-Speech with Flow Matchingcited by 30
  8. 2016The MGB-2 Challenge: Arabic Multi-Dialect Broadcast Media Recognitioncited by 22
  9. 2023ClArTTS: An Open-Source Classical Arabic Text-to-Speech Corpuscited by 14
  10. 2023ArTST: Arabic Text and Speech Transformercited by 14
  11. 2024Mixat: A Data Set of Bilingual Emirati-English Speechcited by 12
  12. 2025SAC: Neural Speech Codec with Semantic-Acoustic Dual-Stream Quantizationcited by 8
  13. 2025NADI 2025: The First Multidialectal Arabic Speech Processing Shared Taskcited by 7
  14. 2025JoyVoice: Long-Context Conditioning for Anthropomorphic Multi-Speaker Conversational Synthesiscited by 7
  15. 2026LEMAS: Large A 150K-Hour Large-scale Extensible Multilingual Audio Suite with Generative Speech Modelscited by 7
  16. 2025M3-TTS: Multi-modal DiT Alignment & Mel-latent for Zero-shot High-fidelity Speech Synthesiscited by 6
  17. 2025ArVoice: A Multi-Speaker Dataset for Arabic Speech Synthesiscited by 4
  18. 2024Towards Zero-Shot Text-To-Speech for Arabic Dialectscited by 3
  19. 2025VietASR: Achieving Industry-level Vietnamese ASR with 50-hour labeled data and Large-Scale Speech Pretrainingcited by 3
  20. 2025DiSTAR: Diffusion over a Scalable Token Autoregressive Representation for Speech Generationcited by 3
  21. 2025Task Vector in TTS: Toward Emotionally Expressive Dialectal Speech Synthesiscited by 3

Built on by

Cited by year (2026 to 14 Oct)

Citing papers per year: 2026: 3.

3 papers.

  1. 2026OmniVoice: Towards Omnilingual Zero-Shot Text-to-Speech with Diffusion Language Modelscited by 25
  2. 2026Dziri Voicebot: An End-to-End Low-Resource Speech-to-Speech Conversational System for Algerian Dialectcited by 0
  3. 2026CARDAMOM: A Micro-Dialectal Arabic Speech Dataset for ASRcited by 0