Children's Speech Recognition Challenge
Organized by Gates Foundation / DrivenData
$70,000 combined prize pool from the Gates Foundation. Build speech recognition systems that accurately transcribe children's voices across languages.
Prize Pool
$70,000
Starts
Sun, Feb 15, 2026
Ends
Sun, Apr 5, 2026
Location
Online
Requirements
What you need to participate
- Free DrivenData account
- Audio/speech processing knowledge
- Two tracks: word ASR and phonetic ASR
- Open to teams and individuals
Tech Stack & Tools
Details
The Gates Foundation is funding two $35,000 challenges on DrivenData to build speech recognition systems that accurately transcribe children's voices — a currently underserved area in ASR technology.
Children's speech has unique characteristics: higher pitch, pronunciation variations, simpler vocabulary, and emotional expression patterns that make standard ASR models less accurate.
The competition features two tracks — word-level ASR and phonetic-level ASR — for a combined $70,000 in prizes. This is meaningful AI for social impact work.