Good with detail? Turn it into flexible remote work AI learns to read documents from examples that people label by hand. Every form, every field, every checkbox a model gets right in the real world traces back to someone who marked it correctly first.
Husky (Speaker) – Native Audio Recording Study
Denmark, Netherlands, Ireland, United Kingdom, Finland, Belgium, Canada, France, Austria, Germany, Switzerland, Norway, Spain
May 13, 2026
Job description
Your voice is exactly what AI needs
AI speech and language models learn from real conversations — not scripts, not recordings of single speakers reading prompts. They need authentic, multi-speaker discussions with natural turn-taking, reactions, laughter, and the rhythm of how people actually talk.
That’s where your natural speaking voice comes in.
You’ll join a small group of 3 native speakers for a natural, unscripted audio discussion — similar to a podcast or talk show episode — on everyday topics provided in advance. As a speaker, your role is to share your thoughts, react naturally, and keep the conversation flowing. Sessions are 30 minutes, fully remote, and paid hourly.
What you’ll do
- Join a 30-minute recorded audio session online with 2 other native speakers
- Participate as a speaker/guest in a natural, unscripted conversation on provided topics
- Share your opinions, stories, and reactions naturally and conversationally — no reading from a script
- A moderator will guide the conversation — your job is to engage, respond, and contribute authentically
- The conversation should feel like a real podcast or talk show — casual, engaging, and authentic
- Submit a brief episode summary after the session
You’ll participate in 3 to 5 sessions total, scheduled at times that work for you.
Why this matters
Your natural speech — the way you react to others, express agreement or disagreement, tell stories, and switch between topics — helps train AI systems to understand real human conversation. This project contributes to making speech recognition and language models work better across languages, dialects, and real-world speaking styles.
How it works
- Apply — select your language and confirm your eligibility
- Get matched — you’ll be grouped with 2 other native speakers
- Review your topics — receive everyday discussion topics in advance
- Join the session — connect online for a 30-minute recorded discussion as a speaker
- Submit your summary — provide a brief episode recap after the session
- Get paid — receive payment for approved recordings
Who can participate
You’re eligible if you:
- Are a native speaker of one of the listed languages
- Are comfortable speaking naturally and spontaneously on everyday topics
- Enjoy sharing your thoughts and engaging in group conversation
- Are available for scheduled 30-minute sessions (3–5 sessions total)
- Have a reliable internet connection and a quiet recording environment
- Have a device with a microphone — headset preferred
What you’ll need
- A device with a microphone (headset preferred for best audio quality)
- A quiet recording environment
- Reliable internet connection
- Availability for 3–5 scheduled sessions
Available regions
Each region below is open to applicants in that country only. Apply through the link for your region.
Country
States / regions
Apply
Apply now
You will be redirected to the company's website to complete your application.
Job scope: data labelling analyst with experience in product matching for catalogs; Brand knowledge/Identify Counterfeit Summary: The main function of a data labeling analyst is to create and manage labeling and change processes within the data management systems.
Put your Korean listening skills to work shaping the future of AI. Native speakers in Sydney wanted — fully remote, flexible, and well paid....
Turn your native Japanese listening skills into AI impact. Transcribe real audio, shape the future of speech technology — fully remote, flexible, and ...