Models / ASAT-v3.1

Speech Synthesis

ASAT-v3.1

Speech across scripts

Speech for Darija and Tamazight across Arabic, Latin, and Tifinagh scripts. ASAT-v3.1 also speaks English, French, Classical Arabic, Spanish, and mixed-language lines.

Explore recorded examples

Open Playground

Version

ASAT-v3.1

Languages

Detected from text

Writing

Arabic · Latin · Tifinagh

Voices

22 available

Access

Sawalni API

Voice studio

Give the line a voice.

Write what ASAT should say56/180

Edit this line, choose a voice, then generate speech.

Voice

Public demo · 20 generations per visitor per hour · 180 characters max

Example recording · Yassine

0:00 / 0:00

This recording uses the original example text. Generate to hear your edits.

Recorded example · asat-v3.1 Explore the live API ↗

Read all 11 recorded examples

Darija · Arabic

speech synthesis

Inputهاد الخدمة كتعاون الناس يفهمو المعلومة بسرعة وبصوت واضح.

Audio recording · 5.0 seconds

Darija · Arabizi

speech synthesis

InputLyoum ghadi nchoufou kifach n9derou n7ellou had lmochkil bshwiya.

Audio recording · 4.1 seconds

Tamazight · Tifinagh

speech synthesis

Inputⴰⵣⵓⵍ ⴼⵍⵍⴰⵡⵏ ⵎⴰⵏⵉ ⵜⴻⵍⵍⴰⵎ ⴰⵙⵙⴰ

Audio recording · 3.6 seconds

Tamazight · Latin

speech synthesis

InputMa ulac aɣilif, awi-d isallen ɣef lḥala n ṭṭaqs n Casablanca kan.

Audio recording · 3.8 seconds

Darija + English

speech synthesis

Inputقبل ما نصيفط التقرير النهائي، كتبت: please verify the final numbers and flag any inconsistency before the review. After that, we should test the full workflow from the user request to the final response. وقلت ليهم إلا كانت شي نقطة غامضة يسولوني دابا.

Audio recording · 16.7 seconds

Darija + French

speech synthesis

Inputباش مايبقاش حتى سوء تفاهم، جاوبتهم: il faut comparer les résultats réels avec les objectifs annoncés au début du projet. Ensuite, la priorité est de vérifier les données avant de prendre une décision définitive. ومن بعد رجعت للدارجة باش نتأكد بلي كلشي فهم.

Audio recording · 17.1 seconds

Darija + Spanish

speech synthesis

Inputفالمكالمة ديال الصباح وضحت ليهم: si la prueba falla, volvemos inmediatamente a la última versión estable. Después, el objetivo principal es mejorar la calidad sin interrumpir el servicio actual. وأكدت بلي هاد القرار خاصو يبقى مكتوب فالمحضر.

Audio recording · 16.3 seconds

English

speech synthesis

InputVirtual teams are held to the same standards of excellence as conventional teams, but there are subtle differences.

Audio recording · 6.9 seconds

French

speech synthesis

InputEn 1990, il a été ajouté à la liste des sites du patrimoine mondial en danger, en raison de la menace des sables du désert.

Audio recording · 7.6 seconds

Spanish

speech synthesis

InputGracias a ella hoy tenemos muchos medios de transporte, como el ferrocarril o el automóvil, entre tantos otros.

Audio recording · 6.6 seconds

Classical Arabic

speech synthesis

Inputمَاسَّ الْحَاجَةِ ظَاهِرَ الْعَجْزِ جَعَلَ لِنَيْلِ حَاجَتِهِ أَسْبَابًا

Audio recording · 3.8 seconds

Model card

About ASAT-v3.1

Languages and writing

Write Darija in Arabic or Latin letters, Tamazight in Tifinagh or Latin letters, or use English, French, Classical Arabic, and Spanish. ASAT-v3.1 reads the text without a language selector. Version 3.1 improves pronunciation in Darija and Classical Arabic.

Voices and pronunciation

The 22 example voices have everyday Moroccan names, such as Yassine and Salma. Each name is also an API alias. Numbered voice IDs still work. Pronunciation and accent can vary, especially in Darija written with Latin letters and numbers or in sentences that mix languages.

Build with asat

Where it fits

See the full stack →

Companion

Speak the answer in Darija or Arabic.

Multilingual media

Give translated or adapted text a voice.

Purpose

Speech across scripts and languages

ASAT-v3.1 accepts Darija in Arabic and Latin scripts, Tamazight in Tifinagh and Latin scripts, and English, French, Classical Arabic, and Spanish. It does not need a language selector. The recorded examples show both single-language and mixed-language speech.

Voice Assistants

Add spoken replies to conversational products and services.

Accessible Content

Turn written information into audio for more inclusive experiences.

Media Localization

Create speech for explainers, lessons, and localized content.

Product Prototyping

Test voice-enabled ideas through one consistent API platform.

Try ASAT-v3.1

Generate multilingual speech with ASAT-v3.1 through the Sawalni API.

Open Speech Playground