Arabic Dialects · Speech to Text

Omani · عُماني

Omani Arabic
Speech to Text

An STT model built for Omani Arabic, understanding Swahili loanwords, Persian vocabulary, and the conserved classical Arabic features that make Omani the most phonologically distinctive of all Gulf dialects.

dialect: "omani"2M+ native speakers · real-time · Muscat, Dhofar, Musandam

2M+

Native Omani speakers

<9%

WER on Omani Arabic speech

<300ms

First-word streaming latency

3

Omani sub-dialects supported

Try it

Speak Omani Arabic.
Watch it transcribe.

Record yourself speaking Omani Arabic and watch the words appear as you talk. The demo runs Hamsa's full production STT, the same model your API calls will use.

Speech to Text · speech in → text out

No sign-up
Microphone
press to record

Your words will appear here.

Press record and speak Arabic or English. The words appear as you talk.

Free to start · in your browserTranscribe in Hamsa Media

Dialect overview

What is Omani Arabic?

Omani Arabic is the most distinctive of the Gulf dialects, carrying the marks of Oman's long history across the Arabian Peninsula, East Africa, and Iran. It is unique in the Gulf for its Swahili influences from the Zanzibar era and its Persian influences from centuries of maritime trade, and for preserving classical Arabic sounds that other Gulf dialects have dropped.

Key features that distinguish Omani Arabic from other Gulf dialects:

  • Classical /ج/: the jim keeps its classical pronunciation, unlike typical Gulf speech.
  • Conserved /ق/: the qaf is preserved in most sub-dialects.
  • Swahili vocabulary: loanwords inherited from the Zanzibar era.
  • Persian vocabulary: in seafaring, ports, and place names.

Native speakers

2M+

Sultanate of Oman

Sub-dialects

3

Muscat · Dhofar · Musandam

Unique influences

Swahili + Persian

Historical linguistic heritage

Hamsa WER

<9%

on real Omani audio · 2024

The problem

Why generic Gulf STT fails
on Omani Arabic

Generic Gulf and MSA-trained models score between 39% and 55% word error rate on real Omani speech. Here is why.

Swahili and Persian influences confuse generic models

Omani Arabic contains Swahili and Persian vocabulary that does not exist in the dictionaries of other Gulf dialects. Generic models transcribe these words incorrectly or drop them entirely, ruining transcript accuracy in half the cases.

Conserved classical sound patterns

Omani Arabic has kept classical Arabic sounds that the rest of the Gulf has dropped, such as the classical /ج/ and the conserved /ق/. Models calibrated on other Gulf dialects misread these sounds.

Three divergent sub-dialects

Muscat urban speech differs fundamentally from southern Dhofar Arabic and from isolated Musandam Arabic in the north. Models built on a single Gulf pattern fail to tell these groups apart and transcribe them with high error rates.

Technical terminology in key sectors

Maritime, oil and gas, and tourism in Oman use a mix of Arabic, English, and Persian technical terms. Generic models that are not calibrated for these sectors produce transcripts that cannot be used.

The solution

How Hamsa handles
Omani Arabic natively

✓

Authentic, multi-source Omani data

Millions of hours of real Omani speech from Muscat, Dhofar, and Musandam, including service calls, tourism content, and professional communications. Not repurposed Gulf data.

✓

A language model with the Omani lexicon

A language model trained on Omani vocabulary, including common Swahili and Persian loanwords and the terminology used in ports, the energy sector, and tourism.

✓

Smart routing across Omani sub-dialects

The model detects dialect features automatically and routes each segment to the best-matched acoustic model (Muscat, Dhofar, or Musandam) with no manual intervention.

✓

Specialist vocabulary for key sectors

Built-in vocabulary lists for oil and gas, ports, tourism, and digital government in Oman, with further customization available for your organization.

Capabilities

Everything you need for
Omani Arabic STT

✓

Native Omani Transcription

Trained on real Omani speech across the Muscat, Dhofar, and Musandam varieties. Not a generic Gulf model re-tuned for Oman.

✓

Swahili and Persian Vocabulary

Omani Arabic carries Swahili loanwords from the Zanzibar era and Persian vocabulary from centuries of trade. Hamsa recognizes these words accurately.

✓

Sub-dialect Coverage

Acoustic models for Muscat, Dhofar, and Musandam speech patterns, with each audio segment routed automatically to the best-matched model.

✓

Real-time Streaming

Under 300ms first-word latency. Works in real time for contact centers, tourism, and maritime services.

✓

Speaker Diarization & Punctuation

Automatic speaker labelling and correct punctuation for Omani Arabic sentence structure.

✓

Custom Vocabulary

Add company, product, and sector terms (oil and gas, ports, tourism) to improve recognition accuracy in your domain.

Transcript examples

Real Omani Arabic.
Accurate transcription.

Hotel & tourism

هيه، الغرفة جاهزة، وباكر الصبح بنوديكم وادي شاب إن شاء الله.

"Yes, the room is ready, and tomorrow morning we'll take you to Wadi Shab, God willing."

Omani: "هيه" (yes), "باكر" (tomorrow), "بنوديكم" (we'll take you)

Port operations

الحاوية ما وصلت لين الحين، باغي أعرف متى بتنزل في ميناء صحار.

"The container hasn't arrived yet. I want to know when it will be unloaded at Sohar Port."

Omani: "لين الحين" (until now), "باغي" (I want), "بتنزل" (will be unloaded)

Customer service

والله اتصلت عليكم أمس وانتظرت واجد، وما حد رد علي.

"Honestly, I called you yesterday and waited a long time, and nobody answered me."

Natural Omani: "واجد" (a lot), "ما حد" (nobody), "رد علي" (answered me)

Use cases

Where Omani Arabic STT
delivers results

Tourism

Omani tourism guest experience

Accurate Omani speech recognition in Muscat, Salalah, and Nizwa tourism and hospitality settings.

Logistics & Maritime

Port communication transcription

Converting operational communications at Port Sultan Qaboos, Port of Sohar, and Port of Salalah to accurate text.

Oil & Gas

PDO report transcription

Converting operational reports and calls in Petroleum Development Oman and the energy sector to text.

Digital Government

Oman digital gateway services

Accurate Omani speech recognition for digital government service platforms.

Accuracy benchmark

High-accuracy STT for Omani Arabic

Measured on a held-out test set of real Omani speech. Lower word error rate means more words transcribed correctly.

Word Error Rate (WER) · lower is better

Hamsa STT (Omani)us9%
Generic Gulf model39.2%
MSA-trained model49.5%
GPT-4o-transcribe54.1%

Evaluated on real Omani 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, suited to telephony platforms, microphones, and recording systems.

02

Dialect routing

Hamsa detects Omani Arabic features automatically and routes audio to the dedicated acoustic and language models, distinguishing between Muscat, Dhofar, and Musandam speech.

03

Transcript out

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

Developer API

Two lines to Omani Arabic speech recognition

Drop in the Hamsa SDK, set dialect: "omani", and start streaming transcripts. No preprocessing pipeline, no generic Gulf fallback.

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

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

stream.on('transcript', (t) => console.log(t.text));
// → "كيف حالك يا أخي؟" · 00:01.840

Related services

More Hamsa for Omani Arabic

Why Hamsa

Built for Arabic from the ground up

✓

Native, not adapted

Hamsa trains on real Omani speaker data, not a generic Gulf model re-tuned for Oman. The difference is 25 to 40 WER points on Omani speech.

✓

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 A/B tests across Omani and other Gulf dialects. No separate integrations, no separate contracts.

FAQ

Omani Arabic STT:
frequently asked

Does the model understand Swahili and Persian vocabulary in Omani Arabic?

Yes. The model's lexicon covers the Swahili- and Persian-origin loanwords common in Omani Arabic, which are missing from the vocabulary of generic Gulf models.

Does the model cover both Muscat and Dhofar Arabic?

Yes. The model is trained on data from both main regions as well as Musandam Arabic, and routes each audio segment automatically to the most suitable acoustic model.

Is the model suitable for tourism and maritime?

Yes. Specialist terminology for tourism, ports, and oil is included in the model, and you can add custom vocabulary for your company name or sector.

What makes Omani Arabic harder to recognize than other Gulf dialects?

Swahili and Persian influences and conserved classical sound patterns make Omani Arabic fundamentally different from the Gulf pattern that generic models are trained on. The variation between Muscat, Dhofar, and Musandam speech adds to the challenge for a single Gulf model.

Can the model be customized with company or sector terminology?

Yes. You can add custom vocabulary lists covering brand names, technical terms, and proper nouns to improve recognition accuracy in your specific domain.

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.