Andromeda Sadiradra - Audio Rater - Generalist - English (Canada)

Welo is an AI services company focused on training and improving voice AI systems. The Welo Data team is building datasets that help make conversational AI agents smarter and more natural-sounding. As an Audio Rater, you'll play a direct role in this work by evaluating how well these agents perform in real conversations.

About Welo: Welo specializes in data collection and quality evaluation for AI training. The company works with organizations across industries to create the datasets needed for machine learning models to improve. Their focus on detailed human feedback helps ensure AI systems become more accurate, helpful, and human-like over time.

Your main task is straightforward: listen to recorded conversations between users and AI voice agents, then rate what you hear. You'll evaluate each agent response on two separate scales, both ranging from 1 to 5. The first scale measures content—whether the response was actually useful, on-topic, and appropriate for the situation. The second scale focuses on prosody—the technical side of speech, including natural tone, pacing, and emotional expression. After scoring, you'll write brief, specific explanations for your ratings. These written justifications help the team understand your reasoning and maintain consistency across evaluations.

This is a freelance position starting as soon as possible, with a commitment of roughly 10 hours per week. You work entirely remotely from Canada and only need fluency in English. The pay is $26.00 per hour.

Who you are matters here. You need strong listening skills and genuine attention to detail—the kind of focus that catches subtle differences in tone and meaning. You should be able to give ratings that make sense and stick with your criteria across multiple evaluations. Your written feedback needs to be clear and direct, not lengthy but precise. You'll also need to show up consistently for the hours you commit each week.

This role suits someone who enjoys working with audio, has patience for careful listening, and wants to contribute to how AI systems learn and improve.