Core Participants
| Role | What They Do | How They Earn Rewards |
|---|---|---|
| Miners | Run voice models, process tasks, output audio | Based on validator quality scores |
| Validators | Send tasks, evaluate outputs, set chain weights | From consistent/accurate scoring |
| Users | Submit real requests (post-mainnet via APIs) | Pay in $TAO for fast/high-quality results |
Task Types
- Synthetic: Auto-generated by validators for consistent benchmarking (edge cases, diverse voices/texts).
- Organic: Real user-submitted jobs—drives adoption and revenue.
Task Format Example
json
{
"task_id": "unique_task_identifier",
"task_type": "synthetic | organic",
"voice_url": "https://storage.example/ref-audio.wav",
"transcript_url": "https://storage.example/target-text.txt",
"transcript_text": "Alternative to transcript_url: direct text content",
"timeout": 60,
"metadata": {
"language": "en",
"duration_hint": 10.5,
"user_id": "optional_user_identifier"
}
}Miner returns hosted audio URL. Full protocol details in whitepaper.