In recent years the increase in the availability of computerized speech recognition and speech synthesis has lead to exciting possibilities in the field of foreign language learning. Computerized speech essentially represents a final stage in the development of a human-computer interface, and in this context, it offers substantial advantages over traditional touch-based interfaces which can eschew language altogether. One activity where this is particularly pertinent is in spaced-learning activities where traditionally students do not need to vocalize their responses, and indeed the responses are often not available to instructors at all. Speech Recognition has further benefits in that every utterance is immediately displayed for students, giving them a more accurate indication of their success with the tested language constructs. The advantage to instructors is in the ability for utterances to be stored as text in a database allowing computer analysis of speech patterns to discern common errors. As the 3rd year of a four-year cross-institutional research grant from the Japanese Government (Kakenhi), this paper will present a speech recognition and speech synthesis system developed by the author within the context of a spaced learning program. We will further show (a) a pattern analysis of the accuracy of the system and patterns of learner usage, (b) an analysis of the effectiveness of spaced-learning using online speaking on student outcomes over 3 institutions, (c) student feedback and reactions on speaking to a machine, (d) how the system deals with pronunciation.