Consonant duration from an Arabic speech database

Rajouani, A.; Ouadou, M.; Nour, Hanna Abdel; Najim, M.

doi:10.1109/melcon.1996.551401

Cited by 1 publication

(3 citation statements)

References 4 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…They were introduced into nonsense words of the structures /TA++ASA/ and /TA++SA/, the plus signs indicate the location of the diphone. These two structures were chosen to cover all the syllabic components of MSA [9]. The recorded nonsense words were then sampled at 16kHz, coded into 16 bits each, and the obtained digital speech was processed by PSOLA speech engine.…”

Section: A-selection Of the Speakermentioning

confidence: 99%

“…Modern Standard Arabic (MSA) is the language commonly understood in all Arab countries. Speech Synthesis of MSA has already been realized using Klatt synthesis [1]. This method uses the formants as its basic building blocks to generate sounds.…”

Section: Introductionmentioning

confidence: 99%

“…Next, the segmentation of the diphones was manually done according to certain rules that were defined by the researchers. These rules took into account the particularities of the MSA [9] and the restrictions imposed by the used synthesizer. With this speaker's voice, a dictionary of diphones was thus produced.…”

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

Text-to-Speech Synthesis by Diphones for Modern Standard Arabic

Nour¹,

Ghattas²

2005

An-Najah University Journal for Research - A

View full text Add to dashboard Cite

An unlimited vocabulary text-to-speech synthesis by diphones system is used to generate Modern Standard Arabic speech: the system is the PSOLA algorithm; the diphones are obtained from the permutation of 44 phones (phonemes and allophones). The diphonic combinations were introduced in carrier words and recorded by a selected speaker. A dictionary of diphones was established by means of a process of segmentation that abided by certain rules. Evaluation of the system was undertaken to assess the accuracy on word and sentence levels. The results showed high perception levels. استُخدم نظام "تركيب الكلام من النص بواسطة ثنائيات الأصوات" لإنتاج كلام باللغة العربية الحديثة دون حدود للمفردات: النظام هو برمجيات PSOLA وحصلنا على ثنائيات الأصوات من تباديل 44 صوتا مختلفا. أدخلت هذه الأزواج على كلمات "ناقلة" وسجلت بصوت قارئ مختار، ومن ثم تم تجزئة هذه التسجيلات حسب قواعد محددة للحصول على "قاموس" من ثنائيات الأصوات. قُيِّم هذا النظام على مستويين: دقة الكلمة ودقة الجملة. دلت النتائج على درجة عالية من الوضوح.

show abstract

Section: A-selection Of the Speakermentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%