Speech & Accent Evaluator — Southern Levantine Arabic (Jordan)

About OpenTrain

OpenTrain is the #1 platform for people who launch and grow careers in AI training and data labeling. We connect contributors with projects that teach AI systems how to speak, read, and understand the world—helping you build experience in a fast-growing field while working remotely and flexibly.

About AI training and this work

AI training (also called data labeling or human feedback) is the human side of building intelligent systems: people provide examples, corrections, and ratings that models learn from. This project focuses on speech — you will evaluate AI-generated audio in Southern Levantine Arabic and help the model speak more naturally and clearly for Jordanian users.

• 100% remote work possible — this project requires you to be located in Jordan for accent familiarity.
• Great part-time flexibility: you can often choose hours to fit studies, work, or family life.

The role

Title: Speech & Accent Evaluator — Southern Levantine Arabic (Jordan). You will interact with an AI system in Southern Levantine Arabic: provide audio prompts, listen to the model’s spoken responses, rate them, and write structured feedback to help engineers and linguists improve pronunciation, fluency, and naturalness.

• Employment type: Contractor, Part-time.
• Workload: Less than 20 hours per week.
• Pay: $20 USD per hour (PAY_PER_HOUR).
• Data type: Audio. Label type: Evaluation rating.
• Location requirement: You must be based in Jordan.

What you'll do day to day

Your core task is to evaluate AI-generated spoken Arabic and provide clear, actionable feedback. Work is rated, iterative, and detail-oriented—every rating and comment helps the model improve.

• Listen to AI audio in Southern Levantine Arabic and assign evaluation ratings for accent, fluency, and pronunciation.
• Provide concise, structured notes identifying errors, unnatural phrasing, or mispronunciations and suggest improvements.
• Create or record short audio prompts when requested to test specific pronunciation or prosody.
• Support iterative refinement by evaluating successive model versions until the output meets quality standards.

Requirements

Candidates must meet the following non-negotiable requirements; do not apply unless you are currently based in Jordan and meet the language experience threshold.

• At least 2 years of daily use of Southern Levantine Arabic (Jordan) in personal or professional communication.
• Currently located in Jordan to ensure familiarity with local pronunciation and accent variation.
• Strong attention to detail and the ability to provide clear, structured feedback on speech.
• Comfort listening to recorded audio and writing concise evaluation comments in either Arabic or English as requested.

Nice-to-have skills

The project welcomes candidates with relevant training, but these are optional and not required to apply.

• Background in linguistics, phonetics, speech analysis, or language teaching.
• Experience evaluating or annotating audio data for ML projects.

Who should apply

This role suits entry-level contributors who are fluent in Southern Levantine Arabic and live in Jordan. You do not need formal qualifications if you meet the language and location requirements and can give detailed, structured feedback.

• Entry-level allowed — ideal for language teachers, linguistics students, bilingual speakers, or anyone with strong local dialect knowledge.
• Good fit for people seeking flexible, part-time remote work contributing directly to how AI systems speak Arabic.

How hiring and the work process work

After you apply, you'll complete a short qualification task to demonstrate your ability to rate and comment on Arabic audio. Qualified contributors are given access to the project's annotation interface (a custom/other platform) and the labeling guidelines. Expect iterative tasks where you evaluate successive model outputs and record notes until the audio reaches quality targets.

• Assessment will check linguistic accuracy, ability to spot accent and fluency issues, iterative improvement mindset, and clarity of feedback.
• You may sometimes be asked to act like an AI interviewer: greeting a candidate, explaining tasks, and running structured evaluations to assess others’ language proficiency.