Acoustic model study device and speech input system

摘要

1st sound environment of the said type every of type of 1st sound environment makes sound occur shows the fluctuation which goodness of fit to the sample voice data of the 1st fluctuation model which and, 2nd sound environment of the said type every of type of 2nd sound environment makes sound occur shows the fluctuation which it integrates goodness of fit to the sample voice data of the 2nd fluctuation model which and with the type of 1st sound environment and does not specify in either of types of 2nd sound environment goodness of fit to the sample voice data of the unspecific acoustic model which integration fitnessIn order to become highest, parameter of the respective model is presumed. Therefore, making use of the sample voice data which receives the environmental influence plural sound the occasion where it constructs the acoustic model, it is possible precision to extract the influence to the sound which originates in the sound environment of husband well.