Job Description

Chemin has partnered with a leading language solutions company to develop advanced AI models and enable human-like conversations for enterprise communication.

We are seeking fluent Cantonese speakers to join our team as remote audio reviewers. This role focuses on reviewing pre-annotated audio files (pre-transcribed, tagged, and segmented). As an Audio Reviewer, you will assess the quality of audio content and make corrections where needed, ensuring high accuracy for AI model development. This role is ideal for detail-oriented individuals who value flexible work hours.

Ultimately, you will help ensure reliable training data and contribute to the development of next-generation AI-powered language tools that accurately represent local speech.

Responsibilities

• Review short audio clips that are already transcribed and tagged
• Correct errors in:
• Transcriptions
• Speaker labels (e.g. gender, role)
• Audio segmentation
• Make sure all spoken content is captured accurately in the correct language(s) (Cantonese/ English / Bahasa Malaysia)
• Perform quality checks to maintain at least 96% accuracy
• Record errors and feedback for improvement
• Communicate clearly with the Project Manager when needed
• Complete tasks on time while maintaining quality standards

 Benefits

• 100% remote work
• RM4.45 per completed task
• With consistent participation, earn up to RM67 per batch
• Training and ongoing project support provided
• Hands-on experience in AI and language data work
• Contribute to AI systems that represent Malaysian languages and accents

Job Requirements

Requirements

• Malaysian or very familiar with Malaysian accents
• Fluent in Cantonese (written) and English
• Basic working knowledge of Bahasa Malaysia (some audios are multilingual)
• Strong listening & comprehension skills, including accents and dialects (Bahasa Malaysia, Cantonese, English)
• High attention to detail and patience
• Comfortable asking questions when unsure
• Laptop or desktop with stable internet
• Nice to have (not required): Experience in transcription, audio review, quality control, or AI data projects
• Must complete and pass a pre-assessment training to be shortlisted. 

Work Scope

• Audio length: around 2–3 minutes per clip
• Time per task: approximately 40–50 minutes
• Main focus: audio review and error identification

Work Arrangements

• Commitment: 4–6 hours per day (weekends may be required)
• Working hours: Flexible
• Join 2 hours of mandatory training that will held between 9:00 am – 6:00 pm (TBD)
• Project period: January - March2026
• Continuous tasks available during the project

Skills