Evaluation device and evaluation method

1. An evaluation device comprising:

a memory storing instructions; and

a processor that implements the instructions to execute a plurality of tasks, including:

an acquisition task that acquires an input sound;

a feature value calculation task that calculates a feature value from the input sound acquired by the acquisition task;

a detection task that detects a break position corresponding to a starting point of each sound included in the input sound acquired by the acquisition task based on the feature value calculated by the feature value calculation task; and

an evaluation value calculation task that calculates, based on time intervals between adjacent break positions among a plurality of break positions detected by the detection task, an evaluation value regarding a degree of temporal regularity of the plurality of break positions.

2. The evaluation device according to claim 1 , wherein:

the feature value includes a value corresponding to a sound-volume level of the input sound, and

the detection task detects the break position further based on a position where the sound-volume level corresponding to the feature value makes a transition from a lower level to a higher level than a predetermined threshold.

3. The evaluation device according to claim 2 , wherein the detection task, to detect the break position, changes the threshold by following a predetermined rule for each predetermined section.

4. The evaluation device according to claim 2 , wherein the sound-volume level is a sound-volume level of the input sound in a predetermined oscillation frequency band.

5. The evaluation device according to claim 1 , wherein:

the feature value includes a value corresponding to an SN ratio of an oscillation frequency distribution of the input sound, and

the detection task detects the break position further based on a position where the SN ratio decreases by a predetermined ratio or greater and then increases.

6. The evaluation device according to claim 1 , wherein:

the feature value includes a value corresponding to an oscillation frequency distribution of the input sound, and

the detection task detects the break position further based on a position where an intensity of a fundamental tone of the oscillation frequency distribution is changed from a lower intensity to a higher intensity than a predetermined threshold.

7. The evaluation device according to claim 1 , wherein:

the feature value includes a value corresponding to a pitch of the input sound, and

the detection task detects the break position further based on a position where a change in the pitch is changed by a predetermined width or greater.

8. The evaluation device according to claim 1 , further comprising:

an accompaniment output task that outputs an accompaniment sound,

wherein the acquisition task acquires the input sound during a period where at least the accompaniment sound is output.

9. An evaluation method comprising:

acquiring an input sound;

calculating a feature value from the acquired input sound;

detecting a break position corresponding to a starting point of each sound included in the acquired input sound based on the calculated feature value; and

calculating, based on time intervals between adjacent break positions among a plurality of detected break points, an evaluation value regarding a degree of temporal regularity of the plurality of detected break positions.

10. The evaluation method according to claim 9 , wherein:

the feature value includes a value corresponding to a sound-volume level of the input sound, and

the detecting of the break position detects the break position further based on a position where the sound-volume level corresponding to the feature value makes a transition from a lower level to a higher level than a predetermined threshold.

11. The evaluation method according to claim 10 , wherein the detecting of the break position, to detect the break position, changes the threshold by following a predetermined rule for each predetermined section.

12. The evaluation method according to claim 10 , wherein the sound-volume level is a sound-volume level of the input sound in a predetermined oscillation frequency band.

13. The evaluation method according to claim 9 , wherein:

the feature value includes a value corresponding to an SN ratio of an oscillation frequency distribution of the input sound, and

the detecting of the break position detects the break position further based on a position where the SN ratio decreases by a predetermined ratio or greater and then increases.

14. The evaluation method according to claim 9 , wherein:

the feature value includes a value corresponding to an oscillation frequency distribution of the input sound, and

the detecting of the break position detects the break position further based on a position where an intensity of a fundamental tone of the oscillation frequency distribution is changed from a lower intensity to a higher intensity than a predetermined threshold.

15. The evaluation method according to claim 9 , wherein:

the feature value includes a value corresponding to a pitch of the input sound, and

the detecting of the break position detects the break position further based on a position where a change in the pitch is changed by a predetermined width or greater.

16. The evaluation method according to claim 9 , further comprising:

outputting an accompaniment sound,

wherein the acquiring of the input sound acquires the input sound during a period where at least the accompaniment sound is output.