Case Study: Leading Multinational Technology Company achieves accurate children's ASR for North American edutainment apps with Appen

Global Tech Firm Expands into New Markets with Enhanced Speech System

Leading Multinational Technology Company needed an automatic speech recognition (ASR) system that could accurately understand children’s speech—something its adult-trained models struggled with due to higher pitch, greater variability and child-specific pronunciations. The company engaged Appen for its language, transcription and speech-recognition expertise to advise on project design and to collect the specialist children’s speech data required.

Appen recruited 400 child speakers across targeted demographics and US dialects, developed education-focused scripts, and collected and transcribed 105 hours (60,000 utterances) of data. Appen’s recommended age-group split (4–7 and 8–14), supervised recording methods and rapid recruitment network helped deliver the dataset on time and on budget, enabling the Leading Multinational Technology Company to build acoustic models and deploy the new North American English ASR across multiple children’s edutainment platforms.