This report presents results obtained in some experiments on the computer recognition of continuous speech. The experiments deal with two simple languages having vocabularies of 11 and 250 words.