Adaptation of Whisper models to child speech recognition