Various types of analyses were applied to digitally recorded speech and musical signals. Resulting parameters were compared, each to the other ones, in order to enable automatic recognition of speech and music patterns. The modified Behrens-Fisher statistic representing the distance between compared elements proved to be an effective tool in the domain of pattern comparison. Results of theoretical approaches and practical experiments are presented in the paper.
This paper costs $33 for non-members and is free for AES members and E-Library subscribers.