Emotion estimation system and non-transitory computer readable medium

1. An emotion estimation system comprising:

a characteristic value extraction unit that analyzes recorded produced speech to extract a predetermined characteristic value;

a vowel section specification unit that specifies, based on the characteristic value extracted by the characteristic value extraction unit, a vowel section in the recorded produced speech, wherein a vowel is produced in the vowel section; and

an estimation unit that estimates, based on whether a characteristic value pattern in the vowel section specified by the vowel section specification unit corresponds to one of types of emotion, an emotion of a speaker, wherein the characteristic value pattern comprises the characteristic value extracted by the characteristic value extraction unit.

2. The emotion estimation system according to claim 1 ,

wherein the estimation unit refers to a vowel probability database in which a characteristic value pattern is recorded for each type of emotion set in advance for each vowel, to obtain a probability that the characteristic value pattern in the vowel section specified by the vowel section specification unit corresponds to each of the types of emotion.

3. The emotion estimation system according to claim 2 ,

wherein the estimation unit divides the vowel section specified by the vowel section specification unit into a plurality of frames, refers to the vowel probability database to obtain a probability that a characteristic value pattern in each of the plurality of frames corresponds to each of the types of emotion, and identifies a type of emotion corresponding to the vowel section, based on the obtained probability for the frame.

4. The emotion estimation system according to claim 3 , further comprising:

a vowel probability database update unit that records characteristic value pattern for each frame in the vowel section as a characteristic value pattern of a vowel corresponding to the vowel section into the vowel probability database.

5. The emotion estimation system according to claim 1 , wherein the estimation unit

estimates, for each vowel section specified by the vowel section specification unit, an emotion of a speaker in the vowel section, based on a characteristic value of the vowel section,

refers to a transition probability database in which a probability that transition from an emotion in a vowel section to an emotion in another vowel section occurs is recorded for each combination of vowels, to obtain a probability of transition of emotion between vowels estimated for each of a plurality of vowel sections included in the produced speech, and

identifies an emotion of the speaker in the entire produced speech, based on a result of estimation of an emotion for each of the vowel sections and the probability of transition of emotion between vowel sections for each combination of vowel sections.

6. The emotion estimation system according to claim 5 , further comprising:

a transition probability database update unit that updates, based on a probability of transition of emotion between vowels for each combination of vowels included in the produced speech obtained by the estimation unit, the probability of transition of emotion between corresponding vowels.

7. A non-transitory computer readable medium storing a program causing a computer to execute a process for emotion estimation, the process comprising:

analyzing recorded produced speech to extract a predetermined characteristic value;

specifying, based on the extracted characteristic value, a vowel section in the recorded produced speech, wherein a vowel is produced in the vowel section; and

estimating, based on whether characteristic value pattern in the vowel section corresponds to one of types of emotion, an emotion of a speaker, wherein the characteristic value pattern comprises the extracted characteristic value.

8. An emotion estimation system comprising:

characteristic value extraction means for analyzing recorded produced speech to extract a predetermined characteristic value;

vowel section specification means for specifying, based on the characteristic value extracted by the characteristic value extraction means, a vowel section in the recorded produced speech, wherein a vowel is produced in the vowel section; and

estimation means for estimating, based on whether a characteristic value pattern in the vowel section specified by the vowel section specification means corresponds to one of types of emotion, an emotion of a speaker, wherein the characteristic value pattern comprises the characteristic value extracted by the characteristic value extraction means.