---
title: "Phonetic Inventory for an Arabic Speech Corpus — Halabi 2016"
description: "The authors tackle a gap in corpus design for Modern Standard Arabic (MSA) speech synthesis. 29 references, cited by 9 papers here."
authors: ["Nawar Halabi", "Mike Wald"]
year: 2016
venue: "Proceedings of the Tenth International Conference on Language Resources and Evaluation (LREC'16)"
source: "https://aclanthology.org/L16-1116/"
licence: "CC BY-NC 4.0 (ELRA, LREC proceedings)"
licence_url: "https://creativecommons.org/licenses/by-nc/4.0/"
summary_licence: "Elamur Cosmic License"
canonical: "https://research.elamur.ai/free/en/papers/halabi-2016-phonetic-inventory-arabic-speech-corpus/"
---
[Elamur Research](https://md.research.elamur.ai/free/en/) · [this page in HTML](https://research.elamur.ai/free/en/papers/halabi-2016-phonetic-inventory-arabic-speech-corpus/)

[home](https://md.research.elamur.ai/free/en/) / [papers](https://md.research.elamur.ai/free/en/papers/) / [2016](https://md.research.elamur.ai/free/en/papers/2016/page/4/) / Phonetic Inventory for an Arabic Speech Corpus

# Phonetic Inventory for an Arabic Speech Corpus

**Authors:** [Nawar Halabi](https://md.research.elamur.ai/free/en/authors/nawar-halabi/) and [Mike Wald](https://md.research.elamur.ai/free/en/authors/mike-wald/) · **Year:** 2016 · **Venue:** [Proceedings of the Tenth International Conference on Language Resources and Evaluation (LREC'16)](https://md.research.elamur.ai/free/en/venues/lrec/) ([LREC 2016](https://md.research.elamur.ai/free/en/venues/lrec/2016/)) · **Publisher:** [European Language Resources Association (ELRA)](https://md.research.elamur.ai/free/en/publishers/european-language-resources-association-elra/) · **Licence:** [CC BY-NC 4.0 (ELRA, LREC proceedings)](https://creativecommons.org/licenses/by-nc/4.0/)

**Links:** [Publisher's page](https://aclanthology.org/L16-1116/) · [PDF](https://aclanthology.org/L16-1116.pdf) · [OpenAlex](https://openalex.org/W2473814316)

**Citations:** 19 citations (OpenAlex, 2026-10-14) · [9](https://md.research.elamur.ai/free/en/papers/halabi-2016-phonetic-inventory-arabic-speech-corpus/cited-by/) in our library, of the 149,439 papers whose bibliographies we read

## Summary
The authors tackle a gap in corpus design for Modern Standard Arabic (MSA) speech synthesis. English has well-studied recording scripts. For MSA, it is unclear which sound units a script must cover. Before choosing sentences, they examine MSA phonetics and phonology to decide which units belong in the "phonetic vocabulary":
- lexical stress on vowels,
- consonant gemination (shadda),
- word-final consonants before pauses,
- intonation constraints.

These decisions become the scoring criteria for a greedy sentence-selection procedure. The source text is a fully diacritised Arabic collection scraped from Aljazeera Learn, which matters because most web Arabic lacks vowel marks. The selection is run several times with different parameters. Each run shrinks the text to a far smaller script with the same phonetic coverage as the original, and that script is the one chosen for recording. The work is one step toward a fully annotated and segmented MSA speech corpus for unit selection synthesis. It also shows how linguistic analysis, rather than raw sentence counts, should decide recording size.

## Key points
- The starting script has 23,531 words in 2092 sentences, taken from Aljazeera Learn because its text carries full diacritics.
- Geminated consonants count as separate phones, so 28 geminate units join the 28 plain consonants. Stressed and unstressed vowels are also distinguished.
- Stress is assigned with Halpern's (2009) ordered MSA rules, with prefixes ignored:
  - a super-heavy final syllable takes stress;
  - otherwise a monosyllable is stressed;
  - otherwise a disyllable is stressed on the penult;
  - otherwise a heavy penult takes stress;
  - otherwise the antepenult is stressed.
- Corpus reduction is NP-hard (2^2092 candidate subsets). The paper therefore uses greedy methods, following the François & Boëffard taxonomy of greedy, spitting, and exchange variants. Each unit must reach a minimum count, set to 3 in their reference.
- Recording guidance:
  - the speaker reads in a neutral declarative style with no contrastive emphasis;
  - the speaker pauses after some short word-final consonants;
  - pitch changes are kept within limits motivated by Kawai et al., where roughly ±0.2 change still sounded acceptable.

## Interesting ideas
- Split a unit into separate inventory entries whenever a feature (stress, length) changes how it is realised, so coverage is measured over feature-qualified units rather than bare symbols.
- Shrink a large pool greedily while holding coverage fixed, and require each unit to appear several times rather than once.
- Express a phonological decision as a short ordered rule cascade where the first matching condition wins, after stripping prefixes.
- Take acceptable-deviation thresholds from published human perception tests instead of guessing them.
- Hold the data in one neutral register, with no emphasis and steady pitch, so that later modelling is not confounded by style variation.

## For Mova

How these ideas could serve Mova, our small open language model — its files, stages and tests:

- **I1** (tokens): Count tagger and morphology coverage over lemma+feature bundles, such as a verb with a specific tense or person, so that unseen combinations stand out as real gaps.
- **I2** (learn): Select silver sentences for costly LLM-teacher review so every UD relation and feature combination appears at least a few times, which cuts review volume without losing variety.
- **I3** (tokens): Write lemma or affix decisions as explicit first-match cascades with normalisation done first, so every output can be traced to the exact rule that fired.
- **I4** (back): Calibrate how far generated or round-tripped text may drift from the source by anchoring the tolerance to human acceptability judgements.
- **I5** (learn): Keep evaluation or seed sets stylistically controlled, then add marked styles like irony or emphasis as a separate labelled slice so their effects can be measured on their own.

*Our summary of the paper, in our own words — not its text. The work belongs to its authors; read it at the links above.*

## Institutions

- [University of Southampton](https://md.research.elamur.ai/free/en/universities/university-of-southampton/) — University, Southampton, United Kingdom · 10 works · 2,495 citations

📚 Shelves (6): [🖥️ Computer Science](https://md.research.elamur.ai/free/en/shelves/computer-science/) · 16,667 works · [🤖 Artificial Intelligence](https://md.research.elamur.ai/free/en/shelves/computer-science/artificial-intelligence/) · 16,194 works · [🎭 Arts and Humanities](https://md.research.elamur.ai/free/en/shelves/arts-and-humanities/) · 16,039 works · [🗣️ Language and Linguistics](https://md.research.elamur.ai/free/en/shelves/arts-and-humanities/language-and-linguistics/) · 15,891 works · [💭 Psychology](https://md.research.elamur.ai/free/en/shelves/psychology/) · 1,552 works · [💭 Experimental and Cognitive Psychology](https://md.research.elamur.ai/free/en/shelves/psychology/experimental-and-cognitive-psychology/) · 870 works

🧭 Topics (2): [Semitic languages](https://md.research.elamur.ai/free/en/topics/semitic-languages/) · 2,908 works · [🎙️ Speech & spoken language](https://md.research.elamur.ai/free/en/topics/speech/) · 480 works

## Cite

Cite as: Nawar Halabi and Mike Wald. 2016. Phonetic Inventory for an Arabic Speech Corpus. Proceedings of the Tenth International Conference on Language Resources and Evaluation (LREC'16).

The paper itself, not this page. BibTeX from the library's record of the work.

```bibtex
@inproceedings{halabi2016phonetic,
  title = {{Phonetic Inventory for an Arabic Speech Corpus}},
  author = {Nawar Halabi and Mike Wald},
  booktitle = {Proceedings of the Tenth International Conference on Language Resources and Evaluation (LREC'16)},
  year = {2016},
  doi = {10.63317/37sgxcd8c2wo},
  url = {https://doi.org/10.63317/37sgxcd8c2wo}
}
```

- [← previous Multi-lingual Dependency Parsing Evaluation: a Large-scale Analysis of Word Order Properties using Artificial Data](https://md.research.elamur.ai/free/en/papers/gulordava-2016-multi-lingual-dependency-parsing-evaluation/)
- [next → The Role of Modifier and Head Properties in Predicting the Compositionality of English and German Noun-Noun Compounds: A Vector-Space Perspective](https://md.research.elamur.ai/free/en/papers/walde-2016-role-modifier-head-properties-predicting/)
- [↑ back to papers of 2016](https://md.research.elamur.ai/free/en/papers/2016/page/4/)

## References

29 references, 1 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

1. Aljazeera. (2015). Aljazeera Learn. Retrieved February 15, 2015, from [http://learning.aljazeera.net/arabic](http://learning.aljazeera.net/arabic) [↗](http://learning.aljazeera.net/arabic)
2. Barros, M., & Möbius, B. (2011). Human Language Technology. Challenges for Computer Science and Linguistics. (Z. Vetulani, Ed.) (Vol. 6562). Berlin, Heidelberg: Springer Berlin Heidelberg. doi:10.1007/978-3-642-20095-3 [↗](https://doi.org/10.1007/978-3-642-20095-3)
3. Biadsy, F., Hirschberg, J., & Habash, N. (2009). [Spoken Arabic dialect identification using phonotactic modeling](https://md.research.elamur.ai/free/en/papers/biadsy-2009-spoken-arabic-dialect-identification-phonotactic/), 53–61. Retrieved from [http://dl.acm.org/citation.cfm?id=1621774.1621784](http://dl.acm.org/citation.cfm?id=1621774.1621784) [↗](http://dl.acm.org/citation.cfm?id=1621774.1621784)
4. Black, A. W. (2002). Perfect Synthesis For All Of The People All Of The Time. In IEEE 2002 Workshop on Speech Synthesis. Retrieved from [http://citeseerx.ist.psu.edu/viewdoc/summary?doi=10.1.1.19.8631](http://citeseerx.ist.psu.edu/viewdoc/summary?doi=10.1.1.19.8631) [↗](http://citeseerx.ist.psu.edu/viewdoc/summary?doi=10.1.1.19.8631)
5. Black, P. E. (2005). Greedy algorithms. Retrieved March 15, 2015, from [http://www.nist.gov/dads/HTML/greedyalgo.html](http://www.nist.gov/dads/HTML/greedyalgo.html) [↗](http://www.nist.gov/dads/HTML/greedyalgo.html)
6. Bonafonte, A., Adell, J., Esquerra, I., Gallego, S., Moreno, A., & Pérez, J. (2008). Corpus and Voices for Catalan Speech Synthesis. In LREC 2008. Retrieved from http://aclweb.org/anthology/L08-1517 [↗](https://aclanthology.org/L08-1517/)
7. Bozkurt, B., Dutoit, T., & Pagel, V. (2002). Re-Defining Intonation From Selected Units For Non-Uniform Units Based Speech Synthesis. In Proc. SPS-IEEE Benelux Signal Process. Symp (pp. 141–144). Leuven, Belgium. Retrieved from [http://citeseerx.ist.psu.edu/viewdoc/summary?doi=10.1.1.19.2935](http://citeseerx.ist.psu.edu/viewdoc/summary?doi=10.1.1.19.2935) [↗](http://citeseerx.ist.psu.edu/viewdoc/summary?doi=10.1.1.19.2935)
8. Bozkurt, B., Dutoit, T., Prudon, R., D’Alessandro, C., & Pagel, V. (2002). Improving quality of MBROLA synthesis for non-uniform units synthesis. In Proceedings of 2002 IEEE Workshop on Speech Synthesis, 2002. (pp. 7–10). IEEE. doi:10.1109/WSS.2002.1224360 [↗](https://doi.org/10.1109/wss.2002.1224360)
9. Buckwalter, T. (2002). Buckwalter Arabic Transliteration. Retrieved June 27, 2013, from [http://www.qamus.org/transliteration.htm](http://www.qamus.org/transliteration.htm) [↗](http://www.qamus.org/transliteration.htm)
10. Clark, R. A. J., Richmond, K., & King, S. (2007). Multisyn: Open-domain unit selection for the Festival speech synthesis system. Speech Communication, 49(4), 317–330. [↗](https://doi.org/10.1016/j.specom.2007.01.014)
11. de Jong, K., & Zawaydeh, B. A. (1999). Stress, duration, and intonation in Arabic word-level prosody. Journal of Phonetics, 27(1), 3–22. doi:10.1006/jpho.1998.0088 [↗](https://doi.org/10.1006/jpho.1998.0088)
12. Essa, O. (1998). Using Prosody in Automatic Segmentation of Speech. In Proceedings of the ACM 36 th Annual Southeast Conference (pp. 44– 49). New York, New York, USA. Retrieved from [http://citeseerx.ist.psu.edu/viewdoc/summary?doi=10.1.1.31.4359](http://citeseerx.ist.psu.edu/viewdoc/summary?doi=10.1.1.31.4359) [↗](http://citeseerx.ist.psu.edu/viewdoc/summary?doi=10.1.1.31.4359)
13. Fernández, E. M., & Cairns, H. S. (2011). Fundamentals of Psycholinguistics (1st ed.). [find](https://scholar.google.com/scholar?q=Fern%C3%A1ndez%2C+E%2E+M%2E%2C+%26+Cairns%2C+H%2E+S%2E+%282011%29%2E+Fundamentals+of+Psycholinguistics+%281st+ed%2E%29%2E)
14. François, H., & Boëffard, O. (2002). The Greedy Algorithm and its Application to the Construction of a Continuous Speech Database. In Proceedings of the Third International Conference on Language Resources and Evaluation (LREC’02). European Language Resources Association (ELRA). Retrieved from http://aclweb.org/anthology/L02-1265 [↗](https://aclanthology.org/L02-1265/)
15. Halpern, J. (2009). Word Stress and Vowel Neutralization in Modern Standard Arabic. In 2nd International Conference on Arabic Language Resources and Tools (pp. 1–7). Cairo, Egypt. [find](https://scholar.google.com/scholar?q=Halpern%2C+J%2E+%282009%29%2E+Word+Stress+and+Vowel+Neutralization+in+Modern+Standard+Arabic%2E+In+2nd+International+Conference+on+Arabic+Language+Resources+and+Tools+%28pp%2E+1%E2%80%937%29%2E+Cairo%2C+Egypt%2E)
16. Kawai, H., Yamamoto, S., Higuchi, N., & Shimizu, T. (2000). A Design Method of Speech Corpus for Text-To-Speech Synthesis Taking Account of Prosody. In Sixth International Conference on Spoken Language Processing (ICSLP 2000) (pp. 420–425). Beijing, China. [find](https://scholar.google.com/scholar?q=Kawai%2C+H%2E%2C+Yamamoto%2C+S%2E%2C+Higuchi%2C+N%2E%2C+%26+Shimizu%2C+T%2E+%282000%29%2E+A+Design+Method+of+Speech+Corpus+for+Text%2DTo%2DSpeech+Synthesis+Taking+Account+of+Prosody%2E+In+Sixth+International+Conference+on+Spoken+Languag)
17. Kawanami, H., Masuda, T., Toda, T., & Shikano, K. (2002). Designing speech database with prosodic variety for expressive TTS system. In Proceedings of the Third International Conference on Language Resources and Evaluation (LREC’02) (pp. 2039–2042). European Language Resources Association (ELRA). Retrieved from http://aclweb.org/anthology/L02-1337 [↗](https://aclanthology.org/L02-1337/)
18. Kelly, A. C., Berthelsen, H., Campbell, N., Ni Chasaide, A., & Gobl, C. (2006). Speech Technology for Minority Languages: the Case of Irish (Gaelic). In INTERSPEECH. Pittsburgh, Pennsylvania. Retrieved from [http://www.tara.tcd.ie/handle/2262/39404](http://www.tara.tcd.ie/handle/2262/39404) [↗](http://www.tara.tcd.ie/handle/2262/39404)
19. Kenworthy, J. (1987). Teaching English Pronunciation. [find](https://scholar.google.com/scholar?q=Kenworthy%2C+J%2E+%281987%29%2E+Teaching+English+Pronunciation%2E)
20. Kominek, J., & Black, A. W. (2003). CMU Arctic Databases for Speech Synthesis. Retrieved from [http://citeseerx.ist.psu.edu/viewdoc/summary?doi=10.1.1.64.8827](http://citeseerx.ist.psu.edu/viewdoc/summary?doi=10.1.1.64.8827) [↗](http://citeseerx.ist.psu.edu/viewdoc/summary?doi=10.1.1.64.8827)
21. Kominek, J., & Black, A. W. (2014). THE CMU ARCTIC SPEECH DATABASES. In 5th ISCA Speech Synthesis Workshop (pp. 223– 224). Pittsburgh, Pennsylvania. [find](https://scholar.google.com/scholar?q=Kominek%2C+J%2E%2C+%26+Black%2C+A%2E+W%2E+%282014%29%2E+THE+CMU+ARCTIC+SPEECH+DATABASES%2E+In+5th+ISCA+Speech+Synthesis+Workshop+%28pp%2E+223%E2%80%93+224%29%2E+Pittsburgh%2C+Pennsylvania%2E)
22. Maia, R., Toda, T., Zen, H., Nankaku, Y., & Tokuda, K. (2007). An Excitation Model for HMM-Based Speech Synthesis Based on Residual Modeling. In 6th ISCA Workshop on Speech Synthesis (pp. 131–136). Bonn, Germany: International Speech Communication Association. Retrieved from [http://library.naist.jp/dspace/handle/10061/8269](http://library.naist.jp/dspace/handle/10061/8269) [↗](http://library.naist.jp/dspace/handle/10061/8269)
23. Matoušek, J., & Romportl, J. (2007). Recording and annotation of speech corpus for Czech unit selection speech synthesis. In TSD’07 Proceedings of the 10th international conference on Text, speech and dialogue (pp. 326–333). Springer-Verlag. Retrieved from [http://dl.acm.org/citation.cfm?id=1776334.1776380](http://dl.acm.org/citation.cfm?id=1776334.1776380) [↗](http://dl.acm.org/citation.cfm?id=1776334.1776380)
24. Newman, D. (1986). The Phonetics of arabic. Journal of the American Oriental Society, 1–6. [find](https://scholar.google.com/scholar?q=Newman%2C+D%2E+%281986%29%2E+The+Phonetics+of+arabic%2E+Journal+of+the+American+Oriental+Society%2C+1%E2%80%936%2E)
25. Oliveira, L., Paulo, S., Figueira, L., Mendes, C., Nunes, A., & Godinho, J. (2008). Methodologies for Designing and Recording Speech Databases for Corpus Based Synthesis. In LREC 2008. Retrieved from http://aclweb.org/anthology/L08-1484 [↗](https://aclanthology.org/L08-1484/)
26. Parlikar, A., & Black, A. W. (2012). Data-driven phrasing for speech synthesis in low-resource languages. In 2012 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP) (pp. 4013–4016). IEEE. doi:10.1109/ICASSP.2012.6288798 [↗](https://doi.org/10.1109/icassp.2012.6288798)
27. Selouani, S.-A., & Caelen, J. (1998). Arabic phonetic features recognition using modular connectionist architectures. In Proceedings 1998 IEEE 4th Workshop Interactive Voice Technology for Telecommunications Applications. IVTTA ’98 (Cat. No.98TH8376) (pp. 155–160). IEEE. doi:10.1109/IVTTA.1998.727712 [↗](https://doi.org/10.1109/ivtta.1998.727712)
28. Tao, J., Liu, F., Zhang, M., & Jia, H. (2008). Design of Speech Corpus for Mandarin Text to Speech. In The Blizzard Challenge 2008 workshop. Brisbane, Australia. [find](https://scholar.google.com/scholar?q=Tao%2C+J%2E%2C+Liu%2C+F%2E%2C+Zhang%2C+M%2E%2C+%26+Jia%2C+H%2E+%282008%29%2E+Design+of+Speech+Corpus+for+Mandarin+Text+to+Speech%2E+In+The+Blizzard+Challenge+2008+workshop%2E+Brisbane%2C+Australia%2E)
29. Tench, P. (2015). Consonants. Retrieved March 15, 2015, from [http://www.cardiff.ac.uk/encap/contactsandpeople/academic/tench/c](http://www.cardiff.ac.uk/encap/contactsandpeople/academic/tench/c) onsonants.html Yi, J. R.-W. (2003). Corpus-based unit selection for natural-sounding speech synthesis. Massachusetts Institute of Technology. Retrieved from http://dspace.mit.edu/handle/1721.1/16944 [↗](http://www.cardiff.ac.uk/encap/contactsandpeople/academic/tench/c)

*Updated 1 November 2026*