Arabic Dialects · Speech to Text

Sudanese · سوداني

Sudanese Arabic
Speech to Text

An STT model for Sudanese Arabic, understanding the distinctive dynamic rhythm and unique vocabulary, serving telecom, banking, and digital wallet applications in Sudan.

dialect: "sudanese"30M+ native speakers · real-time · Khartoum, Nile Valley & Kordofan

30M+

Native Sudanese Arabic speakers

<13%

WER on Sudanese colloquial speech

<300ms

First-word streaming latency

3+

Sudanese sub-dialects covered

Coming Soon

Demo launching soon

The interactive demo for this dialect is coming soon. Sign up to be notified when it launches.

Join the waitlist →

Dialect overview

What is Sudanese Arabic?

Sudanese Arabic is a highly distinctive Arabic variety spoken by more than 30 million people in Sudan. It is characterized by a dynamic rhythm and a vocabulary that blends Classical Arabic with Nilotic and Sudanese African linguistic influences, alongside traces of Hejazi Arabic. Urban Khartoum is the prestige standard, while Nile Valley, Kordofan, and Darfur varieties carry distinct regional features.

Key features that distinguish Sudanese Arabic from Modern Standard Arabic:

  • /ق/ → /g/ · The qaf is a hard G. "قال" (he said) sounds like "gaal".
  • /ث/ and /ذ/ often become /t/ and /z/ or /d/. "تلاتة" (three) instead of "ثلاثة".
  • Distinctly Sudanese vocabulary: "زول" (person), "داير" (I want), "هسع" (now).
  • Everyday expressions: "كيفنك؟" (how are you?), "شنو؟" (what?), "سمح" (good), "عديل" (straight, really).

Native speakers

30M+

Sudan

Sub-dialects

3+

Khartoum · Nile Valley · Kordofan

Streaming latency

<300ms

first word, real-time

Hamsa WER

<13%

on colloquial Sudanese audio · 2024

The problem

Why generic Arabic STT fails
on Sudanese Arabic

Models trained on MSA struggle with everyday Sudanese speech. Here is why.

MSA models do not understand colloquial Sudanese

Most Arabic STT models are trained on formal MSA content, while Sudanese people speak a highly distinctive dialect in daily life, with its own pronunciation, vocabulary, and structures. The result is a transcript that does not reflect what the speaker actually said.

Sudanese vocabulary missing from dictionaries

Sudanese Arabic blends Classical Arabic with Nilotic and Sudanese African linguistic influences. Everyday words like "زول" (person), "داير" (want), and "هسع" (now) are unknown to generic models, which replace them with wrong words or drop them.

Dynamic rhythm and speech rate

Sudanese speech has a dynamic rhythm and a distinctive pace. Models never trained on that rhythm misplace word and sentence boundaries, and transcription accuracy drops.

Regional variation within Sudan

Urban Khartoum is the prestige standard, but Nile Valley, Kordofan, and Darfur varieties keep distinct regional features. A model that is not specialized reads these differences as noise or errors.

The solution

How Hamsa handles
Sudanese Arabic natively

✓

Real Sudanese training data

Models built on data from real Sudanese speakers across multiple regions, including customer service calls and natural conversation. Not repurposed MSA data.

✓

Sudanese language model

A language model trained on actual Sudanese vocabulary and structures, including words of Nilotic and African origin and everyday expressions that never appear in MSA dictionaries.

✓

Acoustic models for Sudanese rhythm

Acoustic models that understand the dynamic rhythm and distinctive pronunciation of Sudanese speech, so words and sentences are segmented accurately even in fast speech.

✓

Khartoum, Nile Valley, and Kordofan coverage

One model covers urban Khartoum speech and the Nile Valley and Kordofan varieties, handling regional differences without separate configuration.

Capabilities

Everything you need for
Sudanese Arabic STT

✓

Native Sudanese Transcription

Trained on real Sudanese speech from multiple regions, not MSA with a dialect layer on top. It understands the dynamic rhythm and Sudanese vocabulary as people actually speak them.

✓

Sub-dialect Coverage

One model handles urban Khartoum speech alongside Nile Valley and Kordofan varieties, with no separate setup for each region.

✓

Speaker Diarization

Identify and label individual speakers in call recordings. Essential for customer service centers at Sudanese telecom operators and banks.

✓

Smart Punctuation

Sentence boundaries, commas, and question marks are inserted automatically following the structure of Sudanese speech, not MSA grammar rules.

✓

Real-time Streaming

Under 300ms first-word latency. Transcript fragments arrive while the speaker is still talking, with no waiting for the end of the sentence.

✓

Custom Vocabulary

Add digital wallet product names, banking services, and your industry's terminology to improve recognition accuracy in your domain.

Transcript examples

Real Sudanese Arabic.
Accurate transcription.

Telecom customer service

يا زول، الشبكة عندي قاطعة من أمبارح وداير أعرف المشكلة شنو.

"Hey, my network has been down since yesterday and I want to know what the problem is."

Sudanese: "زول" (person, mate), "أمبارح" (yesterday), "داير" (want), "شنو" (what)

Digital wallet

داير أحوّل قروش لأخوي هسع، بس التطبيق ما راضي يفتح.

"I want to send money to my brother right now, but the app won't open."

Natural Sudanese: "قروش" (money), "هسع" (now), "ما راضي" (refuses to, won't)

Banking call

كيفنك؟ والله الحساب ده سمح عديل، بس الرسوم كتيرة شوية.

"How are you? Honestly, this account is really good, but the fees are a bit high."

"كيفنك" (how are you), "سمح" (good), "عديل" (really), "كتيرة" (/ث/ → /t/)

Use cases

Where Sudanese Arabic STT
delivers results

Telecom

Telecom customer service call transcription

High-accuracy transcription of Sudanese telecom customer service calls for quality and compliance.

Digital Banking

Banking call documentation

Converting calls from banks, Islamic banking, and agricultural banking to text documents for compliance.

Digital Wallet

Money transfer services

Speech recognition in Sudanese Arabic for digital wallet apps and mobile money transfer services.

Digital Government

Digital citizen services

Accurate Sudanese speech recognition for digital government service platforms.

Accuracy benchmark

High-accuracy STT for Sudanese Arabic

Measured on a held-out benchmark of Sudanese customer service calls and natural conversation. Lower word error rate means more words transcribed correctly.

Word Error Rate (WER) · lower is better

Hamsa STT (Sudanese)us13%
Generic Arabic STT40.2%
MSA-trained model50.3%
GPT-4o-transcribe54.1%

Evaluated on Sudanese colloquial speech benchmarks · 2024 · Lower is better

Architecture

How it works

01

Audio in

Send audio via WebSocket stream or REST upload. Hamsa accepts PCM, MP3, M4A, WAV, and WebM, so it works with contact center systems, mobile apps, and recording platforms.

02

Dialect routing

Hamsa detects the acoustic patterns of Sudanese Arabic and routes audio to its dedicated acoustic and language models, accounting for the differences between Khartoum speech and Nile Valley and Kordofan varieties.

03

Transcript out

Transcripts stream as JSON events with word-level timestamps, speaker labels, confidence scores, and punctuation. Pipe them directly into your database, UI, or downstream NLP pipeline.

Developer API

Two lines to Sudanese Arabic speech recognition

Drop in the Hamsa SDK, set dialect: "sudanese", and start streaming transcripts. No preprocessing pipeline, no normalization to MSA, no fallback surprises.

Read the API docs →
sudanese-stt.tsstreaming
import { Hamsa } from '@hamsa-ai/sdk';

const stream = hamsa.stt.stream({
  model:       'hamsa-stt-s4',
  dialect:     'sudanese',
  diarization: true,
});

stream.on('transcript', (t) => console.log(t.text));
// → "كيفنك يا زول؟ داير أحوّل قروش هسع" · 00:01.840

Related services

More Hamsa for Sudanese Arabic

Why Hamsa

Built for Arabic from the ground up

✓

Native, not adapted

Hamsa trains on real Arabic dialect data, including Sudanese speakers from multiple regions. Not English models adapted for Arabic, and not MSA with a dialect patch.

✓

Enterprise-grade SLA

99.9% uptime, SOC 2 Type II compliance, and data residency in the MENA region. Your audio stays in-region.

✓

Fastest Arabic streaming

Under 300ms first-word latency in production. Real-time transcription that keeps pace with human speech rates.

✓

13+ dialects, one API

Switch dialect with a single parameter. Run Sudanese alongside Egyptian or Gulf with no separate integrations and no separate contracts.

FAQ

Sudanese Arabic STT,
frequently asked

What is the difference between Sudanese Arabic and Modern Standard Arabic?

Modern Standard Arabic is the formal register used in media, education, and official documents. Sudanese Arabic is the spoken language of more than 30 million people in Sudan, with a dynamic rhythm and a vocabulary that blends Classical Arabic with Nilotic and Sudanese African linguistic influences. That is why STT models trained on MSA struggle noticeably with natural Sudanese speech.

Does the model support Sudan's different regional varieties?

Yes. The model is trained on urban Khartoum speech and on Nile Valley and Kordofan varieties, and it handles the phonetic and lexical differences between them without separate configuration.

What word error rate does Hamsa achieve on Sudanese Arabic?

Hamsa achieves under 13% WER on real colloquial Sudanese speech. The gap with generic models is clearest on calls and natural, unscripted conversation, which is exactly the data most enterprises actually need to process.

Is the model suitable for real-time contact centers?

Yes. Real-time streaming with under 300ms latency is designed for customer service centers at telecom operators and banks, with speaker diarization and automatic punctuation.

Which industries use Sudanese Arabic STT most?

The leading ones are telecom, for customer service call transcription; digital banking, Islamic banking, and agricultural banking, for call documentation and compliance; digital wallet apps and money transfer services; and digital government service platforms.

Explore more

13+ Arabic dialects, supported

Partner ecosystem

Trusted Across the
Technology Ecosystem

Hamsa works with leading technology companies, integrators, and solution providers to bring enterprise Voice AI to organizations across the region.