Audio Segmentation and Transcription

About Kaya

Kaya by Chemin Sdn Bhd is a community for high-performing data annotators who play an integral role in shaping the future of machine learning and artificial intelligence.

Kaya offers a collaborative environment where ambitious annotators can thrive. It is a tight-knit community that supports members' professional growth and helps them build a long-term career in data labelling and AI.

Join Kaya and start contributing to impactful AI projects.

About the Role

We are looking for fluent Cantonese speakers and writers who have a healthcare background or are familiar with medical terminology to join a short-term remote project helping train AI language systems.

The recordings in this project are in Cantonese-English (code-switching), meaning speakers naturally switch between Cantonese and English within the same conversation. As such, fluency in both Cantonese and English is required.

Your task in this project is to listen to Cantonese-English audio recordings and transcribe them accurately. Many of these recordings contain healthcare-related conversations, making familiarity with medical terminology essential for producing high-quality transcriptions. Your work will help AI better understand authentic Cantonese-English speech across different speakers, accents, and speaking styles.

If you're comfortable working with audio, detail-oriented, and can commit a few hours a day, this project is for you.

Requirements

What You'll Be Doing?

• Listen to Cantonese-English audio clips and transcribe them accurately
• Segment audio into natural speech units
• Identify and label speakers when multiple voices are present
• Add simple annotations (e.g. gender, age group, emotion)
• Review and validate transcription quality through Quality Control (QC) tasks, if assigned
• Follow project guidelines to maintain high-quality outputs
• Communicate actively with the Project Manager for clarifications
• Complete assigned tasks within the project timeline

Project Details

• Duration: 2 months
• Project Period: July 20 – September 2026
• Time commitment: At least 4 hours per day (including weekends when required)
• Daily target: Minimum 3 tasks per day
• Onboarding Training: Mandatory (TBD, during office hours 9am-6pm GMT+8)
• Working hours: Flexible (tasks can be completed based on the project deadline)

Pay Rate

• Execution: USD 0.10 per accepted task (AHT: ~3–5 minutes per task)

• Quality Control (QC): USD 0.10 per accepted task 

• Execution Details:

• Audio Length: 1–9 seconds

What Do I Need?

• Fluent in spoken Cantonese and able to write in Traditional Chinese
• Fluent in English for project communication and understanding project guidelines
• Cantonese speakers from Hong Kong and Malaysia are welcome to apply
• Healthcare background or familiarity with medical terminology commonly used in hospital or clinical environments
• Familiar with using a Cantonese Jyutping keyboard
• Strong listening and comprehension skills, with the ability to understand different accents, speaking styles, and background noise
• Detail-oriented and patient, able to maintain at least 95% accuracy
• Willing to ask questions when unsure
• Have access to a laptop or desktop computer, headset, and a stable internet connection
• Ability to manage your own time and meet deadlines
• Complete a mandatory pre-assessment as part of the screening process here: https://forms.gle/D7kUzwkSgVWW7MvQ7

Bonus point: Previous experience in transcription, audio annotation, data labelling, linguistics, healthcare-related language work, or AI-related projects.

Benefits

• Work remotely from anywhere
• Be part of an AI project representing authentic Cantonese-English healthcare conversations
• Priority consideration for future opportunities
• Gain hands-on experience in AI language data, transcription, and quality control
• Contribute to building smarter and more inclusive AI language technologies