Teach AI to hear what customers actually mean.
Customer service AI doesn't just need to understand words, it needs to understand tone. The difference between a calm inquiry and a frustrated follow-up changes everything about how a request should be handled, routed, and resolved.
But AI can't learn emotional nuance from text alone. Models need real human voices delivering the same words with different intent, calm versus tense, patient versus impatient. That contrast is exactly what teaches AI to recognize the signals that matter in real customer interactions.
The Emotional Speech Recording Quest is now live on Perle Labs! Visit app.perle.xyz to start earning points.
What This Task Is
The Emotional Speech Recording Quest is a set of 5 audio recording tasks. Each prompt presents a realistic customer or user statement. Your job is to record two versions of the same script, each using a different emotional tone.
These recordings help train AI systems to better understand emotional nuance, intent, and tone in human speech. The same words, delivered differently, teach models to distinguish between a routine request and an escalation.
What You'll Do
Step 1: Read the provided script carefully.
Step 2: Record the first version using a Passive / Neutral Tone — calm, polite, emotionally controlled.
Step 3: Record the second version using a Frustrated / Aggressive Tone — irritated, tense, impatient (do not shout).
Step 4: Submit both recordings to complete the task.
Recording Requirements
- Do not change, add, or remove any words from the script
- Speak clearly and at a natural volume
- Record in a quiet environment
- Ensure both recordings are easy to understand
- Each recording should reflect the intended emotional tone while keeping the wording identical
The contrast between your two recordings is what makes this data valuable, same words, different emotional delivery.
Why It Matters
By contributing to this task, you are:
- Teaching AI to recognize emotional tone and intent in spoken language
- Building enterprise-grade voice datasets with verified human sources
- Improving the accuracy of customer service routing and escalation systems
- Contributing to the sovereign data layer that enterprises and institutions require
This is nuance work that synthetic audio cannot replicate — the human layer that makes voice AI systems reliable.
Task Details
- Points per question: 100
- Number of prompts: 5
- Total recordings: 10 (2 per prompt)
- Total points available: 500
- Project eligibility: 1,000 points
Ready to Get Started?
The Emotional Speech Recording Quest is now live on your Perle Labs dashboard: http://app.perle.xyz
Each recording you submit helps build auditable AI infrastructure for voice systems where understanding emotion isn't optional, not black-box pipelines, but human-verified data with a clear chain of custody.
Stay Connected
Whether you’re returning from beta or joining for the first time, Season 1 is open and ready. Follow along with us on X, Discord, and Telegram to stay connected so you don’t miss what’s coming.
Human-verified. Auditable. Sovereign. Start contributing.
