Child Speech Recognition in Human-Robot Interaction: Problem Solved?

TL;DR

Recent advances in data-driven speech recognition, especially Transformer models like OpenAI Whisper, significantly improve child speech recognition, enabling more effective human-robot interactions despite remaining challenges.

Contribution

This paper demonstrates that modern Transformer-based models substantially enhance child speech recognition performance, showing potential for real-time, autonomous child-robot communication.

Findings

OpenAI Whisper outperforms commercial cloud services.

Structured interactions improve recognition accuracy.

Achieves 60.3% sentence recognition with sub-second latency.

Abstract

Automated Speech Recognition shows superhuman performance for adult English speech on a range of benchmarks, but disappoints when fed children's speech. This has long sat in the way of child-robot interaction. Recent evolutions in data-driven speech recognition, including the availability of Transformer architectures and unprecedented volumes of training data, might mean a breakthrough for child speech recognition and social robot applications aimed at children. We revisit a study on child speech recognition from 2017 and show that indeed performance has increased, with newcomer OpenAI Whisper doing markedly better than leading commercial cloud services. Performance improves even more in highly structured interactions when priming models with specific phrases. While transcription is not perfect yet, the best model recognises 60.3% of sentences correctly barring small grammatical differences, with sub-second transcription time running on a local GPU, showing potential for usable autonomous child-robot speech interactions.