Abstract: Context-dependent phone models are applied to speaker-independent continuous speech recognition and shown to be effective in this domain. Several previously proposed context-dependent models ...