JP6657887B2 - Voice interaction method, voice interaction device, and program - Google Patents

Description

  The present invention relates to a technology of voice interaction for reproducing a response voice to an utterance voice.

  2. Description of the Related Art A speech dialogue technology for realizing a dialogue with a user by reproducing a voice of a response (for example, an answer to a question) to an utterance by the user has been proposed. For example, Patent Literature 1 discloses a technique of analyzing utterance content by voice recognition of a user's uttered voice, and synthesizing and reproducing a response voice according to the analysis result.

JP 2012-128440 A

  However, under the existing technology including Patent Literature 1, it is actually difficult to realize a natural spoken dialogue that faithfully reflects the tendency of a real human-dialogue conversation, and it is mechanically unnatural. There is a problem that a user can feel an impression. In view of the above circumstances, the present invention aims at realizing a natural voice dialogue.

  In order to solve the above problems, a voice interaction device according to a preferred aspect of the present invention is a device for executing a voice interaction for reproducing a response voice to an utterance voice, and a voice for acquiring an utterance signal representing the utterance voice. An acquisition unit, a history management unit that generates a usage history of voice dialogue, and a response generation unit that causes a playback device to reproduce a prosody response voice corresponding to the usage history. In the above embodiment, since the response voice of the prosody corresponding to the usage history of the voice dialogue is reproduced, the tendency of the real dialogue that the prosody of the spoken voice changes with time as the dialogue with the specific dialogue partner is repeated is considered. It is possible to realize a simulated natural speech dialogue.

  In a preferred aspect of the present invention, the response generation unit controls a standby time, which is an interval between the speech sound and the response sound, according to the usage history. In the above embodiment, the standby time, which is the interval between the uttered voice and the response voice, is controlled according to the usage history. A natural spoken dialogue simulating the tendency of a real dialogue in which the interval between the dialogues is shortened as it is repeated is realized.

It is a lineblock diagram of a voice dialogue device of a 1st embodiment. 5 is a flowchart of an operation of the voice interaction device according to the first embodiment. FIG. 3 is an explanatory diagram of a speech sound and a response sound in the first embodiment. FIG. 3 is an explanatory diagram of a speech sound and a response sound in the first embodiment. 5 is a flowchart of a response generation process according to the first embodiment. It is a lineblock diagram of a speech dialogue device of a 2nd embodiment. It is explanatory drawing of the speech sound and response sound in 2nd Embodiment. It is explanatory drawing of the speech sound and response sound in 2nd Embodiment. It is a flow chart of response generation processing in a 2nd embodiment. It is a lineblock diagram of a voice interaction device of a 3rd embodiment. It is a flowchart of operation | movement of the voice interaction apparatus in 3rd Embodiment. It is a flow chart of response generation processing in a 3rd embodiment. It is explanatory drawing of the speech sound and response sound in 3rd Embodiment. It is explanatory drawing of the speech sound and response sound in 3rd Embodiment. It is a lineblock diagram of a voice dialogue device of a 4th embodiment. It is a flowchart of operation | movement of the voice interaction apparatus in 4th Embodiment. It is a flow chart of response generation processing in a 4th embodiment. It is explanatory drawing of the speech sound and response sound in 4th Embodiment. It is explanatory drawing of the speech sound and response sound in 4th Embodiment.

<First embodiment>
FIG. 1 is a configuration diagram of a voice interaction device 100A according to the first embodiment of the present invention. The voice interaction device 100A of the first embodiment is a voice interaction system that reproduces a voice (hereinafter, referred to as "response voice") Vy in response to a voice (hereinafter, referred to as "uttered voice") Vx pronounced by the user U. For example, a portable information processing device such as a mobile phone or a smartphone, or an information processing device such as a personal computer can be used as the voice interaction device 100A. It is also possible to realize the voice interaction device 100A in the form of a toy (for example, a doll such as a stuffed animal) or a robot that simulates the appearance of an animal or the like.

  The utterance voice Vx is, for example, an utterance voice including a question (question) and a utterance, and the response voice Vy is a response voice including an answer to the question and an acknowledgment to the utterance. The response voice Vy also includes, for example, voices that mean interjections. Interjections are independent words (inflection, exclamation) that are used independently of other segments and have no inflection. To be more specific, words such as “Ye” and “Ee” representing the utterance of the utterance, words such as “Eto” and “Ano” representing the stagnation (stagnation of the response), and the response (affirmation / Negative) words such as "yes" and "no", words such as "oh" and "oh" that express the excitement of the speaker, or "eh?" ? "And the like can be exemplified as interjections.

  The voice interaction device 100A of the first embodiment generates a response voice Vy of a prosody according to the prosody of the utterance voice Vx. Prosody is a linguistic and phonetic characteristic that speech listeners can perceive, and cannot be grasped solely from a general notation of the language (for example, notation excluding special notation indicating prosody). Means nature. The prosody can be rephrased as a characteristic that allows the listener to recall or guess the intention or emotion of the speaker. Specifically, intonation (change in tone of speech, intonation), tone (pitched or intense), pitch (speech length), speech speed, rhythm (structure of temporal change in tone), accent (pitched Although various features such as a strong or weak accent can be included in the concept of prosody, typical examples of prosody are pitch (fundamental frequency) or volume.

  As illustrated in FIG. 1, the voice interaction device 100A according to the first embodiment includes a control device 20, a storage device 22, a voice input device 24, and a playback device 26. The voice input device 24 is, for example, an element that generates a voice signal X (hereinafter, referred to as a “voice signal”) representing the voice Vx of the user U, and includes a sound collection device 242 and an A / D converter 244. The sound collection device (microphone) 242 collects the utterance sound Vx generated by the user U and generates an analog sound signal representing a sound pressure fluctuation of the utterance sound Vx. The A / D converter 244 converts the audio signal generated by the sound collection device 242 into a digital speech signal X.

  The control device 20 is an arithmetic processing device (for example, a CPU) that controls each element of the voice interaction device 100A. The control device 20 of the first embodiment acquires the speech signal X supplied from the speech input device 24, and generates a response signal Y indicating a response speech Vy for the speech speech Vx. The reproduction device 26 is an element that reproduces the response voice Vy according to the response signal Y generated by the control device 20, and includes a D / A converter 262 and a sound emission device 264. The D / A converter 262 converts the digital response signal Y generated by the control device 20 into an analog audio signal, and the sound emitting device 264 (for example, a speaker or headphones) outputs a response audio corresponding to the converted audio signal. Vy is emitted as a sound wave. The playback device 26 may also include a processing circuit such as an amplifier that amplifies the response signal Y.

  The storage device 22 stores a program executed by the control device 20 and various data used by the control device 20. For example, a known recording medium such as a semiconductor recording medium or a magnetic recording medium, or a combination of a plurality of recording media can be arbitrarily adopted as the storage device 22. The storage device 22 of the first embodiment stores an audio signal Z representing a response audio of a specific utterance content. In the following description, a case where the audio signal Z of the response audio such as “un” which means Aizuchi which is an example of the interjection is stored in the storage device 22 will be exemplified. The audio signal Z is recorded in advance and stored in the storage device 22 as an audio file in any format such as a wav format.

  The control device 20 executes a program stored in the storage device 22 to perform a plurality of functions (a voice acquisition unit 32, a voice analysis unit 34A, and a response generation unit 36A) for establishing a dialogue with the user U. Realize. In addition, a configuration in which the function of the control device 20 is realized by a plurality of devices (that is, a system), or a configuration in which a part of the function of the control device 20 is shared by a dedicated electronic circuit may be adopted.

  The speech acquisition unit 32 in FIG. 1 acquires a speech signal X representing the speech voice Vx. The voice acquisition unit 32 of the first embodiment acquires the speech signal X generated by the voice input device 24 from the voice input device 24. The voice analysis unit 34A specifies the pitch (basic frequency) P of the utterance voice Vx from the utterance signal X acquired by the speech acquisition unit 32. The specification of the pitch P is sequentially performed at a predetermined cycle. That is, the pitch P is specified for each of a plurality of different time points on the time axis. A publicly known technique can be arbitrarily adopted for specifying the pitch P of the uttered voice Vx. Note that the pitch P can be specified by extracting an acoustic component of a specific frequency band from the utterance signal X. The frequency band to be analyzed by the voice analysis unit 34A is variably set in accordance with, for example, an instruction from the user U (for example, designation of a male / female voice). Further, the frequency band to be analyzed can be dynamically changed according to the pitch P of the uttered voice Vx.

  The response generation unit 36A causes the reproduction device 26 to reproduce the response voice Vy corresponding to the utterance voice Vx of the utterance signal X acquired by the speech acquisition unit 32. Specifically, the response generation unit 36A generates a response signal Y of the response voice Vy in response to the pronunciation of the utterance voice Vx by the user U, and supplies the response signal Y to the playback device 26, thereby supplying the response voice Vy. On the playback device 26. The response generation unit 36A of the first embodiment adjusts the prosody of the voice signal Z stored in the storage device 22 according to the pitch P of the uttered voice Vx specified by the voice analysis unit 34A, thereby obtaining the response voice Vy. Is generated. That is, the reproduction device 26 reproduces the response voice Vy obtained by adjusting the initial response voice represented by the voice signal Z in accordance with the prosody of the speech voice Vx.

  In a real human dialogue, the conversation partner pronounces a response voice to the utterance voice at a pitch corresponding to the pitch near the end point of the utterance voice of the speaker (that is, the response voice has a pitch of the utterance voice). (Depending on the pitch near the end point). In consideration of the above tendency, the response generation unit 36A of the first embodiment adjusts the pitch of the voice signal Z in accordance with the pitch P of the uttered voice Vx specified by the voice analysis unit 34A, thereby obtaining the response voice. A response signal Y of Vy is generated.

  FIG. 2 is a flowchart of a process executed by the control device 20 of the first embodiment. For example, the process of FIG. 2 is started by an instruction from the user U to the voice interaction device 100A (for example, an instruction to start a voice interaction program).

  When the process of FIG. 2 is started, the voice acquiring unit 32 waits until the user U starts to generate the uttered voice Vx (S10: NO). Specifically, the voice acquisition unit 32 sequentially specifies the volume of the voice Vx by analyzing the voice signal X supplied from the voice input device 24, and determines the volume of the voice Vx to be a predetermined threshold (for example, If the state exceeding the fixed value selected in (1) or the variable value according to the instruction from the user U) continues for a predetermined time length, it is determined that the uttered voice Vx has started. The method of detecting the start of the uttered voice Vx (that is, the start point of the uttered section) is arbitrary. For example, when the volume of the uttered voice Vx exceeds the threshold value and the voice analyzing unit 34A detects a significant pitch P, it is possible to determine that the uttered voice Vx has started.

  When the uttered voice Vx starts (S10: YES), the voice acquiring unit 32 acquires the uttered signal X from the voice input device 24 and stores it in the storage device 22 (S11). The voice analysis unit 34A specifies the pitch P of the utterance voice Vx from the utterance signal X acquired by the speech acquisition unit 32 and stores the pitch P in the storage device 22 (S12).

  The voice acquisition unit 32 determines whether the user U has finished uttering the uttered voice Vx (S13). Specifically, the voice acquisition unit 32 determines that the volume of the utterance voice Vx specified from the utterance signal X is a predetermined threshold (for example, a fixed value selected in advance or a variable value according to an instruction from the user U). When the lowering state continues for a predetermined time length, it is determined that the uttered voice Vx has ended. However, a known technique can be arbitrarily adopted for detecting the end of the uttered voice Vx (that is, the end point of the uttered section). As understood from the above description, during the speech period during which the speech of the speech voice Vx is continued (S13: NO), the acquisition of the speech signal X by the speech acquisition unit 32 (S11) and the speech speech Vx by the speech analysis unit 34A. (S12) of the pitch P is repeated.

  As a result of the processing described above, as illustrated in FIGS. 3 and 4, a time series of a plurality of pitches P of the uttered voice Vx is specified for the uttered section from the start point to the end point tB of the uttered voice Vx. In FIG. 3, it is assumed that the user U has pronounced the utterance voice Vx of the question sentence "Fun?" In which the speaker asks the recognition of the emotion and intention of the utterer. In FIG. 4, it is assumed that the user U has pronounced an utterance voice Vx of a declarative sentence that expresses recognition of the speaker's own feelings and intentions and requests the speaker to give consent to the recognition.

  When the uttered voice Vx ends (S13: YES), the response generator 36A executes a process SA (hereinafter referred to as "response generating process") SA for causing the reproducing device 26 to reproduce the response voice Vy corresponding to the uttered voice Vx. As described above, the response generation processing SA of the first embodiment adjusts the pitch of the voice signal Z in accordance with the pitch P of the uttered voice Vx specified by the voice analysis unit 34A, thereby forming the response signal of the response voice Vy. This is a process for generating Y.

  FIG. 5 is a flowchart of a specific example of the response generation processing SA. As described above, the response generation processing SA in FIG. 5 is started upon the end of the uttered voice Vx (S13: YES). When the response generation process SA is started, the response generation unit 36A determines, as illustrated in FIGS. 3 and 4, a section E (hereinafter, referred to as a “tail section”) including the end point tB of the speech voice Vx among the speech sounds Vx. The lowest value (hereinafter, referred to as "lowest pitch") Pmin of the plurality of pitches P specified by the voice analysis unit 34A is specified as the prosody of the uttered voice Vx (SA1). The end section E is, for example, a part of the utterance voice Vx that extends from the end point tB of the utterance voice Vx to a predetermined length on the near side (for example, several seconds). As can be understood from FIG. 3, the pitch P of the utterance voice Vx of the question sentence tends to increase near the end point tB. Therefore, the pitch P at the minimum point where the transition of the pitch P of the uttered voice Vx changes from a drop to a rise is specified as the minimum pitch Pmin. On the other hand, as understood from FIG. 4, the pitch P of the declarative utterance voice Vx tends to decrease monotonically toward the end point tB. Therefore, the pitch P at the end point tB of the uttered voice Vx is specified as the lowest pitch Pmin.

  The response generator 36A generates a response signal Y representing a response voice Vy having a pitch corresponding to the lowest pitch Pmin of the uttered voice Vx (SA2). Specifically, as illustrated in FIGS. 3 and 4, the response generation unit 36 </ b> A has the lowest pitch at a specific time point (hereinafter referred to as “target point”) τ in the response voice Vy on the time axis. By adjusting the pitch of the audio signal Z so as to match the height Pmin, a response signal Y of the response voice Vy is generated. A preferred example of the target point τ is a start point of a specific mora (typically, the last mora) among a plurality of mora constituting the response voice Vy. For example, assuming the voice signal Z of the response voice "Y", as understood from FIGS. 3 and 4, the pitch at the start point of the last mora "N" of the voice signal Z is the minimum pitch Pmin. The response signal Y of the response voice Vy is generated by adjusting (pitch shifting) the pitch over the entire section of the voice signal Z so as to coincide with the following. A known technique can be arbitrarily adopted for adjusting the pitch. The target point τ is not limited to the start point of the last mora in the response voice Vy. For example, the pitch can be adjusted with the start point and the end point of the response voice Vy as the target point τ.

  When the response signal Y is generated according to the above procedure, the response generation unit 36A waits until a point in time at which reproduction of the response sound Vy should be started (hereinafter referred to as a "response reproduction point") ty (SA3: NO). The response reproduction point ty is, for example, a point in time when a predetermined time (for example, 150 ms) has elapsed from the end point tB of the uttered voice Vx.

  When the response reproduction point ty arrives (SA3: YES), the response generation unit 36A reproduces the response sound Vy by supplying the response signal Y adjusted according to the minimum pitch Pmin to the reproduction device 26 (SA4). . That is, the reproduction of the response voice Vy is started at the response reproduction point ty at which a predetermined time has elapsed from the end point tB of the utterance voice Vx. The response generator 36A sequentially supplies the response signal Y from the response reproduction point ty to the reproducing device 26 in real time in parallel with the generation (pitch shift) of the response signal Y to reproduce the response voice Vy. It is also possible. As understood from the above description, the response generation unit 36A of the first embodiment is configured as an element for causing the playback device 26 to play the response voice Vy having a pitch corresponding to the lowest pitch Pmin in the last section E of the utterance voice Vx. Function.

  When the response generation processing SA illustrated above is completed, the control device 20 determines whether or not the user U has instructed the end of the voice interaction as illustrated in FIG. 2 (S14). If the end of the voice interaction has not been instructed (S14: NO), the process proceeds to step S10. That is, triggered by the start of the uttered voice Vx (S10: YES), the voice obtaining unit 32 obtains the utterance signal X (S11), the voice analyzing unit 34A specifies the pitch P (S12), and the response generating unit 36A And a response generation process SA is performed. As understood from the above description, the response voice Vy having a pitch corresponding to the pitch P of the uttered voice Vx is reproduced for each pronunciation of the uttered voice Vx. In other words, a voice dialogue is realized in which the sounding of an arbitrary uttered voice Vx by the user U and the reproduction of a response voice Vy (for example, a response voice of “un”) of Aizuchi to the uttered voice Vx are alternately repeated. . When the end of the voice conversation is instructed by the user U (S14: YES), the control device 20 ends the processing of FIG.

  As described above, in the first embodiment, the response voice Vy having a pitch corresponding to the lowest pitch Pmin in the last section E including the end point tB of the uttered voice Vx is reproduced from the reproducing device 26. Therefore, it is possible to realize a natural voice dialogue that simulates a tendency of an actual dialogue in which a conversation partner produces a response voice at a pitch corresponding to a pitch near the end point of the uttered voice. In the first embodiment, in particular, the response voice Vy is reproduced such that the pitch at the start point (target point τ) of the last mora of the response voice Vy matches the minimum pitch Pmin. The effect of realizing a natural voice conversation is particularly remarkable.

<Modification of First Embodiment>
(1) In the first embodiment, the configuration in which the pitch of the target point τ in the response voice Vy is made to coincide with the lowest pitch Pmin in the last section E of the utterance voice Vx has been described, but the target point τ of the response voice Vy The relationship between the pitch and the minimum pitch Pmin of the uttered voice Vx is not limited to the above example (a relationship in which both match). For example, the pitch at the target point τ of the response voice Vy can be made to match the pitch obtained by adding or subtracting a predetermined adjustment value (offset) δp to the minimum pitch Pmin. The adjustment value δp is a fixed value selected in advance (for example, a numerical value corresponding to a pitch such as 5 degrees with respect to the minimum pitch Pmin) or a variable value according to an instruction from the user U. Further, according to the configuration in which the adjustment value δp is set to a numerical value corresponding to an integral multiple of the octave, the response voice Vy having the pitch shifted by the octave from the lowest pitch Pmin is reproduced. Whether to apply the adjustment value δp can be switched according to an instruction from the user U.

(2) In the first embodiment, the pitch of the response voice Vy is controlled according to the pitch P of the uttered voice Vx (specifically, the minimum pitch Pmin of the tail section E). The type of prosody of the uttered voice Vx used for control and the type of prosody of the response voice Vy controlled in accordance with the prosody of the uttered voice Vx are not limited to the pitch. For example, a configuration in which the prosody of the response voice Vy is controlled in accordance with the volume (an example of the prosody) of the utterance voice Vx, or the response voice Vy in accordance with the range of the pitch or volume fluctuation of the utterance voice Vx (an example of the prosody) A configuration for controlling the prosody of the sound is also adopted. Further, a configuration for controlling the volume (an example of a prosody) of the response voice Vy according to the prosody of the utterance voice Vx, a range of the pitch or volume fluctuation of the response voice Vy according to the prosody of the speech voice Vx (other than the prosody) For example, a configuration for controlling (e.g.) may be adopted.

(3) In real human dialogue, the prosody of the response voice is not necessarily determined uniformly according to the prosody of the uttered voice. That is, there is a tendency that the prosody of the response voice depends on the prosody of the uttered voice and can vary for each pronunciation of the uttered voice. In consideration of the above tendency, the response generation unit 36A can change the prosody (for example, pitch and volume) of the response voice Vy reproduced from the reproduction device 26 for each utterance voice Vx. Specifically, in the configuration in which the pitch of the response voice Vy is adjusted so that the pitch is obtained by adding or subtracting the adjustment value δp to or from the minimum pitch Pmin, as in the above-described modification, the response generation unit 36A The adjustment value δp is variably controlled for each sound of the voice Vx. For example, the response generation unit 36A generates a random number within a predetermined range for each sound of the uttered voice Vx, and sets the random number as the adjustment value δp. According to the above configuration, it is possible to realize a natural voice dialogue that simulates the tendency of a real dialogue in which the prosody of the response voice can vary for each pronunciation of the utterance voice.

(4) In the first embodiment, the response signal Y is generated by adjusting the pitch of one type of audio signal Z. However, a plurality of types of audio signals Z having different pitches are used for generating the response signal Y. It is also possible. For example, a configuration in which the response signal Y is generated by adjusting the pitch of the audio signal Z closest to the lowest pitch Pmin of the uttered voice Vx among the plurality of types of audio signals Z can be assumed.

(5) In the first embodiment, the response voice Vy is reproduced from the reproduction device 26. However, the utterance voice Vx is reproduced from the reproduction device 26 by supplying the utterance signal X acquired by the audio acquisition unit 32 to the reproduction device 26. It is possible. A configuration in which whether or not the uttered voice Vx is reproduced from the reproducing device 26 is switched according to an instruction from the user U may be adopted.

<Second embodiment>
A second embodiment of the present invention will be described. Note that in the following embodiments, elements having the same functions and functions as those of the first embodiment will be denoted by the same reference numerals used in the description of the first embodiment, and detailed description thereof will be omitted as appropriate.

  FIG. 6 is a configuration diagram of the voice interaction device 100B according to the second embodiment of the present invention. The voice interaction device 100B of the second embodiment reproduces a response voice Vy to the utterance voice Vx pronounced by the user U, similarly to the voice interaction device 100A of the first embodiment. As illustrated in FIG. 6, the voice interaction device 100B of the second embodiment has a configuration in which the response generation unit 36A of the voice interaction device 100A of the first embodiment is replaced with a response generation unit 36B. The configuration and operation of the other elements of the voice interaction device 100B (the voice input device 24, the playback device 26, the voice acquisition unit 32, and the voice analysis unit 34A) are the same as in the first embodiment.

  In a real-life dialogue between humans, a tendency is observed that a conversation partner produces a response voice in a prosody corresponding to the content of the speaker's utterance (whether it is a question sentence or a declarative sentence). For example, the prosody differs between the response voice for the question text and the response voice for the declarative text. Specifically, the voice of the answer to the question sentence is relatively large when compared with the voice of Aizuchi to the declarative sentence, for example, because it is necessary for the speaker to clearly recognize the answer (affirmation / denial) of the respondent. It tends to be pronounced with emphasis (volume or pitch fluctuation over time) at the volume. In consideration of the above tendency, the response generation unit 36B of the second embodiment causes the reproduction device 26 to reproduce the prosody response voice Vy according to the utterance content (discrimination of question text / declaration text) by the utterance voice Vx.

  FIG. 7 illustrates the transition of the pitch P of the utterance voice Vx of the question sentence, and FIG. 8 illustrates the transition of the pitch P of the utterance voice Vx of the declarative sentence. As can be understood from FIGS. 7 and 8, the transition of the pitch P in the vicinity of the end of the utterance voice Vx (temporal change) between the case where the utterance content of the utterance voice Vx is a question sentence and the case where the utterance is a declarative sentence (Trend of fluctuation). Specifically, as illustrated in FIG. 7, the pitch P of the utterance voice Vx of the question sentence changes from a drop to an increase or monotonically increases in the end section E, but the sound of the utterance voice Vx of the declarative sentence The height P monotonously decreases from the start point tA to the end point tB of the end section E as illustrated in FIG. Therefore, by analyzing the transition of the pitch P near the end of the uttered voice Vx (end section E), it is possible to estimate whether the uttered content of the uttered voice Vx corresponds to a question sentence or a declarative sentence. It is.

  In consideration of the above tendency, the response generation unit 36B of the second embodiment generates the response voice of the prosody corresponding to the transition of the pitch P in the last section E of the utterance voice Vx (that is, the distinction between question text / declaration text). The playback device 26 plays back Vy. Specifically, as illustrated in FIG. 7, when the transition of the pitch P of the uttered voice Vx changes from decreasing to rising in the tail section E, or the pitch P of the uttered voice Vx is monotonous in the tail section E. (That is, when the utterance content is presumed to be a question sentence), the playback device 26 reproduces the response voice Vy of a prosody suitable for the question sentence. On the other hand, as illustrated in FIG. 8, when the pitch P of the uttered voice Vx monotonously decreases in the tail section E (that is, when the uttered content is estimated to be a declarative sentence), it is suitable for the declarative sentence. A response voice Vy having a proper prosody is reproduced from the reproduction device 26.

  As illustrated in FIG. 6, the storage device 22 of the voice interaction device 100B of the second embodiment stores a response signal YA and a response signal YB in which a response voice Vy of a specific utterance content is recorded in advance. The response signal YA and the response signal YB have mutually common utterance contents (character notation) but different prosody. Specifically, the response voice Vy represented by the response signal YA is a voice of "Ye" pronounced with the intention of a positive answer to the utterance voice Vx of the question sentence, and the response voice Vy represented by the response signal YB is This is the sound of "Yeah" pronounced by Aizuchi for the utterance voice Vx of the declarative sentence. Specifically, the response voice Vy of the response signal YA has a higher volume than the response voice Vy of the response signal YB, and has a difference in prosody that the range of fluctuation of the volume and the pitch (that is, intonation) is wide. The response generation unit 36B of the second embodiment selectively supplies either one of the response signal YA and the response signal YB stored in the storage device 22 to the playback device 26, so that a plurality of responses having different prosody are provided. The voice Vy is selectively reproduced. Note that it is also possible to make the sound content of the response signal YA different from that of the response signal YB.

  FIG. 9 is a flowchart of a response generation process SB for causing the reproduction device 26 to reproduce the response voice Vy by the response generation unit 36B of the second embodiment. In the second embodiment, the response generation process SA of FIG. 2 exemplified in the first embodiment is replaced with the response generation process SB of FIG. The processes other than the response generation process SB are the same as those in the first embodiment. The response generation processing SB of FIG. 9 is started with the end of the uttered voice Vx (S13: YES).

  When the response generation processing SB is started, the response generation unit 36B calculates an average of a plurality of pitches P (hereinafter, referred to as “first average pitch”) Pave1 in the first section E1 of the last section E of the uttered voice Vx, The average of a plurality of pitches P in the second section E2 (hereinafter referred to as "second average pitch") Pave2 is calculated (SB1). As illustrated in FIGS. 7 and 8, the first section E1 is a preceding section of the end section E (for example, a section including the start point tA of the end section E), and the second section E2 is a section of the end section E. The section is a section behind the first section E1 (for example, a section including the end point tB of the end section E). Specifically, the first half of the last section E is defined as a first section E1, and the second half of the last section E is defined as a second section E2. However, the conditions of the first section E1 and the second section E2 are not limited to the above examples. For example, a configuration in which the first section E1 and the second section E2 move back and forth with an interval, and a configuration in which the first section E1 and the second section E2 have different time lengths may be adopted.

  The response generator 36B compares the first average pitch Pave1 of the first section E1 with the second average pitch Pave2 of the second section E2, and determines whether the first average pitch Pave1 is lower than the second average pitch Pave2. It is determined whether or not (SB2). As described above, the transition of the pitch P of the utterance voice Vx of the question sentence tends to change from decreasing to increasing or to increase monotonously in the last section E. Therefore, as illustrated in FIG. 7, the first average pitch Pave1 is likely to be lower than the second average pitch Pave2 (Pave1 <Pave2). On the other hand, the pitch P of the utterance voice Vx of the declarative sentence tends to decrease monotonously in the last section E. Therefore, as illustrated in FIG. 8, the first average pitch Pave1 is likely to exceed the second average pitch Pave2 (Pave1> Pave2).

  In consideration of the above tendency, if the first average pitch Pave1 is lower than the second average pitch Pave2 (SB2: YES), that is, if the utterance voice Vx is highly likely to be a question sentence, the second The response generator 36B of the embodiment selects the response signal YA corresponding to the response voice Vy of the answer to the question sentence from the storage device 22 (SB3). On the other hand, if the first average pitch Pave1 exceeds the second average pitch Pave2 (SB2: NO), that is, if the utterance voice Vx is highly likely to be a declarative sentence, the response generation unit 36B will The response signal YB corresponding to the response voice Vy of the consent to is selected from the storage device 22 (SB4).

  When the response signal Y (Y1, Y2) corresponding to the transition of the pitch P of the uttered voice Vx is selected in the above procedure, the response generator 36B, like the first embodiment, arrives at the response reproduction point ty (SB5). : YES), the response signal Y is supplied to the reproducing device 26 to reproduce the response voice Vy (SB6). Specifically, when the pitch P of the uttered voice Vx changes from decreasing to rising in the last section E, or when the pitch P of the uttered voice Vx monotonically increases in the last section E (SB2: YES). The response voice Vy of the answer to the question sentence is reproduced, and when the pitch P of the utterance voice Vx monotonously decreases in the last section E (SB2: NO), the response voice Vy of the consent to the declarative sentence is reproduced. . That is, the prosody of the response voice Vy reproduced from the reproducing device 26 differs between the case where the utterance voice Vx is a question sentence and the case where the utterance voice Vx is a declarative sentence.

  The acquisition of the utterance signal X by the voice acquisition unit 32 (S11), the specification of the pitch P by the voice analysis unit 34A (S12), and the response generation process SB by the response generation unit 36B are described as follows. (S14: NO). Therefore, as in the first embodiment, a voice dialogue is realized in which the sound of an arbitrary uttered voice Vx by the user U and the reproduction of a response voice Vy to the uttered voice Vx are alternately repeated.

  As described above, in the second embodiment, the reproduction device 26 reproduces the prosody response voice Vy corresponding to the transition of the pitch P in the tail section E of the utterance voice Vx. Therefore, it is possible to realize a natural voice dialogue that simulates a tendency of a real dialogue in which a conversation partner pronounces a response voice with a prosody corresponding to the content of the utterance of the speaker. In the second embodiment, in particular, the case where the transition of the pitch P changes from a decrease to a rise within the end section E, or the case where the pitch P monotonously increases within the end section E, and the case where the end point E starts from the start point tA. Since the prosody of the response voice Vy is different between the case where the pitch P monotonously decreases toward tB, the natural voice dialogue simulating the tendency of the actual dialogue in which the prosody of the response voice differs between the question sentence and the declarative sentence. Can be realized.

  In the second embodiment, the prosody of the response voice Vy is determined according to the result of comparing the first average pitch Pave1 in the first section E1 of the last section E with the second average pitch Pave2 in the second section E2. Therefore, there is an advantage that the transition of the pitch P can be evaluated (and the prosody of the response voice Vy can be selected) by a simple process of averaging and comparing a plurality of pitches P.

<Modification of Second Embodiment>
(1) In the second embodiment, the plurality of response signals Y (YA, YB) previously stored in the storage device 22 are selectively supplied to the playback device 26. However, a single response signal recorded in advance is used. By adjusting Y, the response generation unit 36B can generate a prosody response signal Y corresponding to the transition of the pitch P in the tail section E of the uttered voice Vx. For example, assuming a configuration in which the response signal YA of the response voice Vy to the declarative sentence is stored in the storage device 22, when the utterance voice Vx is a question sentence, the response generator 36B increases the volume of the response signal YA and increases the volume. While the response signal YB of the response voice Vy of the answer is generated by expanding the range of the variation of the pitch and the pitch, the response signal YA is supplied to the reproducing device 26 when the utterance voice Vx is a descriptive sentence. It is also possible to generate the response signal YA of the response voice Vy consenting to the descriptive sentence by reducing the volume of the initial response signal Y and reducing the range of the volume and pitch fluctuations.

  According to the configuration in which response signals Y having different prosody are generated by adjusting one response signal Y, it is not necessary to store a plurality of response signals Y (YA, YB) having different prosody in the storage device 22, There is an advantage that the storage capacity required for the storage device 22 is reduced. On the other hand, according to the configuration of the second embodiment that selectively uses the plurality of response signals Y having different prosody, it is not necessary to adjust the initial prosody of the response signal Y according to the utterance content of the utterance voice Vx. Therefore, there is an advantage that the processing load on the response generation unit 36B is reduced.

(2) In the second embodiment, of the last section E, the first average pitch Pave1 in the first section E1 and the second average pitch Pave2 in the second section E2 are compared. The method for estimating whether the content corresponds to a question sentence or a declarative sentence is not limited to the above example. For example, since the pitch P monotonously decreases in the end section E of the uttered voice Vx of the declarative sentence, the pitch P tends to be the lowest pitch Pmin at the end point tB of the end section E. Therefore, if the time length of the rear section of the last section E at which the pitch P becomes the minimum pitch Pmin is sufficiently shorter than the preceding section (for example, when the time length falls below a predetermined threshold), the utterance voice It is also possible to estimate that the utterance content of Vx corresponds to the declarative sentence. It is also possible to estimate whether the utterance content of the utterance voice Vx corresponds to a question sentence or a declarative sentence according to the transition of the pitch P before and after the lowest pitch Pmin in the last section E. is there. For example, when the pitch P rises after the lapse of the lowest pitch Pmin in the last section E, the response generation unit 36B estimates that the uttered content of the uttered voice Vx corresponds to the question sentence.

<Third embodiment>
FIG. 10 is a configuration diagram of a voice interaction device 100C according to the third embodiment of the present invention. The voice interaction device 100C of the third embodiment reproduces a response voice Vy to the utterance voice Vx pronounced by the user U, similarly to the voice interaction device 100A of the first embodiment. In the third embodiment, in addition to the answer to the utterance voice Vx or the response voice of Aizuchi (hereinafter, referred to as “second response voice”) Vy2, the response voice representing the inquiry to the utterance voice Vx (hereinafter, referred to as “first response voice”). ) Vy1 can be played from playback device 26. The first response voice Vy1 is a voice such as “What?” Or “What?” For re-listening the utterance voice Vx to the speaker. As illustrated in FIG. 10, the storage device 22 of the voice interaction device 100C according to the third embodiment stores a response signal Y1 containing the first response voice Vy1 of the query and a response signal Y1 other than the query (eg, “Yes”). And a response signal Y2 containing the second response voice Vy2 of the hammer.

  As illustrated in FIG. 10, the voice interaction device 100C according to the third embodiment includes the voice analysis unit 34A and the response generation unit 36A of the voice interaction device 100A according to the first embodiment, and the voice analysis unit 34C and the response generation unit 36C. This is a configuration with replacement. The configuration and operation of the other components of the voice interaction device 100C (the voice input device 24, the playback device 26, and the voice acquisition unit 32) are the same as in the first embodiment.

  The voice analysis unit 34C of the third embodiment specifies the prosody index value Q from the speech signal X acquired by the speech acquisition unit 32. The prosody index value Q is an index value relating to the prosody of the utterance voice Vx, and is calculated for each utterance voice Vx (for each unit when a series of utterances from the start point to the end point of the utterance voice Vx is used as a unit). Specifically, the average value of the pitch, the variation range of the pitch, the average value of the volume, or the variation range of the volume in the utterance section of the uttered voice Vx is calculated from the utterance signal X as the prosody index value Q. As described above, the response generation unit 36C of the third embodiment selectively causes the reproducing device 26 to reproduce the first response sound Vy1 representing the inquiry to the utterance sound Vx and the second response sound Vy2 other than the inquiry.

  In a real human dialogue, when the prosody of a speaker's uttered voice fluctuates, there is a tendency that it becomes difficult for a conversation partner to hear the uttered voice, and the necessity of questioning increases. Specifically, when the prosody of the uttered voice of the speaker deviates from the past prosody tendency of the speaker (for example, the volume of the actual uttered voice is lower than the volume assumed by the conversation partner from the past tendency) In this case), the conversation partner cannot properly hear the uttered voice, and as a result, there is a high possibility that a question is returned to the speaker. In consideration of the above tendency, the response generation unit 36C of the third embodiment compares the prosody index value Q specified by the voice analysis unit 34C with the threshold QTH, and according to the comparison result, the first response voice Vy1 and the One of the two response voices Vy2 is reproduced by the reproducing device 26. The threshold value QTH is set to a representative value (for example, an average value) of the prosody index value Q of the uttered voice Vx uttered by the user U in the past. That is, the threshold QTH corresponds to a standard prosody estimated from the past utterance of the user U. When the prosody index value Q of the uttered voice Vx deviates from the threshold value QTH, the first response voice Vy1 of the query is reproduced, and when the prosody index value Q approximates the threshold value QTH, the second response of Aizuchi is obtained. The sound Vy2 is reproduced.

  FIG. 11 is a flowchart of a process executed by the control device 20 according to the third embodiment. For example, the process of FIG. 11 is started by an instruction from the user U to the voice interaction device 100C (for example, an instruction to start a voice interaction program).

  As in the first embodiment, when the uttered voice Vx is started (S20: YES), the voice obtaining unit 32 obtains the uttered signal X from the voice input device 24 and stores it in the storage device 22 (S21). The voice analysis unit 34C specifies the feature q regarding the prosody of the voice Vx from the voice signal X acquired by the voice acquisition unit 32 (S22). The feature quantity q is, for example, the pitch P or volume of the uttered voice Vx. The acquisition of the speech signal X by the speech acquisition unit 32 (S21) and the specification of the feature q by the speech analysis unit 34C (S22) are repeated until the end of the speech voice Vx (S23: NO). That is, a time series of a plurality of feature quantities q of the uttered voice Vx is specified for the uttered section from the start point to the end point tB of the uttered voice Vx.

  When the uttered voice Vx ends (S23: YES), the voice analyzing unit 34C calculates the prosodic index value Q from a time series of a plurality of feature amounts q specified for the utterance section from the start point to the end point of the uttered voice Vx (S24). ). Specifically, the voice analysis unit 34C calculates, as the prosody index value Q, the average value or the fluctuation range (range) of the plurality of feature amounts q in the utterance section.

  When the prosody index value Q of the current uttered voice Vx is calculated by the process described above, the response generation unit 36C executes a response generation process SC for causing the reproduction device 26 to reproduce the response voice Vy. The response generation process SC of the third embodiment is a process of causing the playback device 26 to selectively reproduce one of the first response voice Vy1 and the second response voice Vy2 according to the prosody index value Q calculated by the voice analysis unit 34C. It is.

When the response generation process SC is completed, the voice analysis unit 34C updates the threshold value QTH according to the prosody index value Q of the current utterance voice Vx (S25). Specifically, the voice analysis unit 34C calculates a representative value (for example, an average value or a median value) of a plurality of prosodic index values Q of the past voice voice Vx including the current voice voice Vx as the updated threshold QTH. . For example, as expressed by the following equation (1), a weighted average (exponential moving average) of the current prosody index value Q and the threshold value QTH before update is calculated as the threshold value QTH after update. The symbol α in Expression (1) is a predetermined positive number (a forgetting factor) smaller than 1.
QTH = α · Q + (1−α) QTH …… (1)
As understood from the above description, the voice analyzing unit 34C of the third embodiment functions as an element that sets the representative value of the prosodic index value Q in the past plurality of uttered voices Vx as the threshold QTH. The threshold value QTH is updated to a numerical value reflecting the prosody index value Q of the uttered voice Vx for each pronunciation of the uttered voice Vx, and becomes a numerical value corresponding to a standard prosody estimated from a plurality of utterances of the user U. . However, the threshold value QTH can be fixed to a predetermined value. For example, the average value of the prosody index values Q specified from the uttered voices of an unspecified number of speakers can be set as the threshold QTH.

  Acquisition of the utterance signal X by the speech acquisition unit 32 (S21), calculation of the prosody index value Q by the speech analysis unit 34C (S22, S24), response generation processing SC by the response generation unit 36C, and threshold by the speech analysis unit 34C. The updating of QTH (S25) is repeated for each sound of the uttered voice Vx until the end of the voice dialogue is instructed by the user U (S26: NO). Accordingly, a voice dialogue in which the pronunciation of the uttered voice Vx by the user U and the selective reproduction of the first response voice Vy1 (interrogation) and the second response voice Vy2 (Aizuchi) are alternately realized. .

  FIG. 12 is a flowchart of the response generation process SC of the third embodiment. When the response generation process SC is started, the response generation unit 36C compares the prosody index value Q specified by the voice analysis unit 34C with the threshold QTH at the current stage, and a predetermined range including the threshold QTH (hereinafter, referred to as “permissible range”). It is determined whether the prosody index value Q is included in R (SC1). FIGS. 13 and 14 illustrate the transition of the feature quantity q specified by the voice analysis unit 34C from the uttered voice Vx. As illustrated in FIGS. 13 and 14, the allowable range R is a range having a predetermined width with the threshold value QTH being a median value. The process (SC1) of comparing the prosody index value Q and the threshold value QTH is to determine whether the absolute value of the difference between the prosody index value Q and the threshold value QTH exceeds a predetermined value (for example, half the range width of the allowable range R). It can also be realized as a determination process.

  In FIG. 13, it is assumed that the prosody index value Q is a numerical value inside the allowable range R. The fact that the prosody index value Q is included in the allowable range R means that the prosody of the current uttered voice Vx approximates the standard prosody of the user U (the tendency of past utterance). That is, assuming a dialogue between real humans, it can be evaluated as a situation in which the conversation partner can easily listen to the uttered voice (a situation in which it is unlikely that a question to the speaker needs to be returned). Therefore, when the prosody index value Q is a numerical value inside the allowable range R (SC1: YES), the response generation unit 36C stores the response signal Y2 of the second response voice Vy2 of the companion voice to the utterance voice Vx from the storage device 22. Select (SC2).

  On the other hand, FIG. 14 assumes that the prosody index value Q is a numerical value outside the allowable range R (specifically, a numerical value lower than the lower limit value of the allowable range R). The fact that the prosody index value Q is not included in the allowable range R means that the prosody of the current uttered voice Vx deviates from the standard prosody of the user U. That is, assuming a dialogue between real humans, it can be evaluated as a situation where it is difficult for the conversation partner to hear the uttered voice (a situation in which it is highly likely that a question to the speaker needs to be returned). Therefore, when the prosody index value Q is a numerical value outside the allowable range R (SC1: NO), the response generator 36C responds to the second response sound Vy1 (eg, "E?" ) Is selected from the storage device 22 as a supply target to the reproducing device 26 (SC3).

  When the response signal Y (the response voice Vy to be reproduced) corresponding to the prosody index value Q is selected in the above procedure, the response generation unit 36C, like the first embodiment, arrives at the response reproduction point ty (SC4: YES). ), The response signal Y is supplied to the reproducing device 26 to reproduce the response voice Vy (the first response voice Vy1 or the second response voice Vy2) (SC5). That is, if the prosody index value Q is included in the allowable range R, the second response voice Vy2 of Aizuchi is reproduced, and if the prosody index value Q is not included in the allowable range R, the first response voice of the inquiry is returned. Vy1 is reproduced.

  As described above, in the third embodiment, the first response voice Vy1 indicating a query to the utterance voice Vx and the second response voice Vy2 other than the query are selectively reproduced from the playback device 26. Therefore, it is possible to realize a natural voice dialogue that simulates the tendency of a real dialogue in which not only a companion to the speaker but also a question (re-listening) to the speaker occurs appropriately.

  In the third embodiment, either the first response voice Vy1 or the second response voice Vy2 is selected according to the result of comparing the prosody index value Q representing the prosody of the voice speech Vx with the threshold QTH. It is possible to realize a natural spoken dialogue that simulates the tendency of a real dialogue that listening becomes difficult and the necessity of questioning increases when the prosody of the speech changes suddenly. Particularly in the third embodiment, since the representative value of the prosody index value Q over a plurality of past speech sounds Vx is set as the threshold QTH, the prosody of the speech sound of the speaker is the standard prosody of the speaker (that is, the dialogue of the speaker). There is also an advantage that a natural spoken dialogue that simulates the tendency of an actual dialogue in which a question from the dialogue partner is likely to occur when deviating from the prosody assumed by the partner is realized. Moreover, when the prosody index value Q is a numerical value outside the allowable range R including the threshold value QTH, the first response voice Vy1 is selected, and when the prosody index value Q is a numerical value inside the allowable range R, the second response voice Vy2 is selected. Therefore, for example, the first response voice Vy1 is excessively frequently generated as compared with a configuration in which the first response voice Vy1 and the second response voice Vy2 are selected only according to the magnitude relation between the prosody index value Q and the threshold value QTH. Can be reduced (the first response voice Vy1 is reproduced at an appropriate frequency).

<Modification of Third Embodiment>
In the third embodiment, the reproduction of the first response voice Vy1 and the reproduction of the second response voice Vy2 are selected in accordance with the prosody index value Q of the voice speech Vx, but the predetermined frequency is determined regardless of the characteristics of the voice speech Vx. It is also possible to reproduce the first response voice Vy1 of the inquiry. Specifically, the response generation unit 36C reproduces the first response voice Vy1 which is interrogated with respect to the utterance voice Vx randomly selected from the plurality of utterance voices Vx that the user U sequentially pronounces, on the reproduction device 26. On the other hand, the second response voice Vy2 of Aizuchi is reproduced for the remaining utterance voice Vx. For example, the response generation unit 36C generates a random number within a predetermined range for each sound of the uttered voice Vx, and selects the first response voice Vy1 when the random number exceeds the threshold, while the random number falls below the threshold. In this case, the second response voice Vy2 is selected. In the modified example described above, since the first response voice Vy1 of the query is reproduced for the speech voice Vx selected at random from the plurality of speech voices Vx, the query for the speech voice is generated randomly. It is possible to realize a natural spoken dialogue simulating the tendency of a real spoken dialogue.

  In the above configuration, the response generation unit 36C can variably set a ratio of the number of times of reproduction of the first response voice Vy1 to the number of times of voice of the utterance voice Vx (that is, the reproduction frequency of the first response voice Vy1). For example, the response generation unit 36C controls the reproduction frequency of the first response sound Vy1 by adjusting the threshold value to be compared with the random number. For example, when the reproduction frequency of the first response voice Vy1 is set to 30%, the first response voice Vy1 is reproduced for 30% of the total number of utterances of the utterance voice Vx, and the remaining 70% The second response voice Vy2 is reproduced in response to the utterance. The reproduction frequency of the first response voice Vy1 (for example, a threshold value to be compared with a random number) is variably set according to an instruction from the user U, for example.

<Fourth embodiment>
FIG. 15 is a configuration diagram of the voice interaction device 100D according to the fourth embodiment of the present invention. The voice interaction device 100D of the fourth embodiment reproduces a response voice Vy corresponding to the utterance voice Vx pronounced by the user U, similarly to the voice interaction device 100A of the first embodiment.

  As illustrated in FIG. 15, the voice interaction device 100D according to the fourth embodiment includes a voice analysis unit 34A and a response generation unit 36A of the voice interaction device 100A according to the first embodiment, and a history management unit 38 and a response generation unit 36D. This is a configuration with replacement. The configuration and operation of the other elements of the voice interaction device 100D (the voice input device 24, the playback device 26, and the voice acquisition unit 32) are the same as in the first embodiment. The storage device 22 of the fourth embodiment stores a response signal Y representing a response voice Vy of a specific utterance content. In the following description, a response voice Vy of "yes" indicating a companion to the utterance voice Vx is exemplified.

  The history management unit 38 in FIG. 15 generates a history (hereinafter, referred to as “usage history”) H of the voice conversation by the voice conversation device 100D. The usage history H of the fourth embodiment is the number N of voice conversations executed in the past using the voice conversation device 100D (hereinafter, referred to as “the number of times of use”). Specifically, one time from the start (activation of the voice interaction apparatus 100D) to the end of the voice interaction (that is, one voice interaction including a plurality of pairs of the utterance voice Vx and the response voice Vy). The history management unit 38 counts the number of voice conversations as the number of uses N. The usage history H generated by the history management unit 38 is stored in the storage device 22.

  The response generation unit 36D of the fourth embodiment causes the reproduction device 26 to reproduce the response voice Vy of the prosody corresponding to the use history H generated by the history management unit 38. That is, the prosody of the response voice Vy is variably controlled according to the usage history H. In the fourth embodiment, the standby time W for reproducing the response voice Vy is controlled as the prosody of the response voice Vy according to the usage history H. The standby time W is a time length from the end point tB of the speech sound Vx to the response reproduction point ty of the response sound Vy (that is, the interval between the speech sound Vx and the response sound Vy).

  In real human dialogue, it is observed that the prosody of the uttered voice changes over time with repetition of the dialogue with a specific partner. More specifically, at the stage immediately after the start of the conversation in the first meeting (the stage in which each person is not used to the conversation with the conversation partner), it is not possible for both parties to grasp a suitable interval specific to the conversation partner. There is a tendency that the time from the utterance to the response to the utterance is long (that is, the conversation is awkward), and the time is shortened (that is, the conversation can be performed at a good tempo) as the conversation with the conversation partner is repeated. In consideration of the above tendency, the response generation unit 36D of the fourth embodiment sets the standby time W of the response voice Vy when the number of uses N indicated by the use history H is larger than when the number of uses N is small. The standby time W is controlled according to the usage history H so as to be shorter.

  FIG. 16 is a flowchart of a process executed by the control device 20 of the fourth embodiment. For example, the processing in FIG. 16 is started by an instruction from the user U (an instruction to start a program for voice dialogue) to the voice dialogue apparatus 100D. At the stage where the voice dialogue by the voice dialogue device 100D is first started, the usage history H is set to an initial value (for example, N = 0).

  As in the first embodiment, when the uttered voice Vx is started (S30: YES), the voice acquiring unit 32 acquires the uttered signal X from the voice input device 24 and stores it in the storage device 22 (S31). The acquisition of the utterance signal X by the voice obtaining unit 32 is repeated until the end of the utterance voice Vx (S32: NO).

  When the uttered voice Vx ends (S32: YES), the response generation unit 36D performs a response generation process SD for causing the reproduction device 26 to reproduce the prosody response voice Vy corresponding to the use history H stored in the storage device 22. Execute. As described above, the response generation process SD of the fourth embodiment is a process of controlling the standby time W from the end point tB of the utterance voice Vx to the response reproduction point ty at which the reproduction of the response voice Vy starts, according to the use history H. It is. The acquisition of the utterance signal X by the voice acquisition unit 32 (S31) and the response generation processing SD by the response generation unit 36D are repeated until the end of the voice interaction is instructed by the user U (S33: NO). Therefore, as in the first embodiment, a voice dialogue is realized in which the sound of an arbitrary uttered voice Vx by the user U and the reproduction of a response voice Vy to the uttered voice Vx are alternately repeated.

  When the end of the voice dialogue is instructed by the user U (S33: YES), the history management unit 38 updates the usage history H stored in the storage device 22 to the content including the current voice dialogue (S34). ). Specifically, the history management unit 38 increases the number of uses N indicated by the use history H by one. Therefore, the usage history H increases by one each time the voice dialogue device 100D executes the voice dialogue. After the use history H is updated, the processing in FIG. 16 ends.

  FIG. 17 is a flowchart of the response generation processing SD of the fourth embodiment, and FIGS. 18 and 19 are explanatory diagrams of the response generation processing SD. When the response generation process SD is started, the response generation unit 36D variably sets the standby time W according to the use history H stored in the storage device 22 (SD1 to SD3). Specifically, the response generation unit 36D first determines whether or not the number of uses N indicated by the use history H exceeds a predetermined threshold NTH (SD1). When the number of uses N exceeds the threshold value NTH (SD1: YES), the response generation unit 36D sets a predetermined basic value w0 (for example, 150 ms) as the standby time W as illustrated in FIG. 18 (SD2). On the other hand, if the number of uses N is less than the threshold value NTH (SD1: NO), the response generator 36D adds a predetermined adjustment value (offset) δw to the base value w0, as illustrated in FIG. 19 (w0 + δw). Is set as the standby time W (SD3). The adjustment value Δw is set to a predetermined positive number. In the above description, the standby time W is binary-controlled according to whether or not the number of uses N exceeds the threshold value NTH. However, the standby time W is multi-valued or continuously controlled according to the number of uses N. It can be changed.

  The response generation unit 36D waits until the standby time W set in accordance with the use history H in the above processing elapses from the end point tB of the uttered voice Vx (SD4: NO). When the response reproduction point ty arrives after the elapse of the standby time W (SD4: YES), the response generation unit 36D reproduces the response sound Vy by supplying the response signal Y stored in the storage device 22 to the reproduction device 26. (SD5). As understood from the above description, the response generation unit 36D of the fourth embodiment reproduces the response voice Vy of the prosody (the standby time W in the fourth embodiment) according to the usage history H of the voice interaction device 100D. To play. Specifically, when the number of uses N indicated by the use history H is large, the response voice Vy is reproduced after the elapse of the standby time W of the base value w0, and when the number of uses N is small, the response voice Vy is adjusted to the base value w0. The response voice Vy is reproduced after the elapse of the standby time W to which the value δw is added. That is, when the number of uses N is large, the standby time W becomes short.

  As described above, in the fourth embodiment, since the response voice Vy of the prosody (standby time W) according to the usage history H of the voice dialogue by the voice dialogue device 100D is reproduced, the dialogue with the specific partner is repeated. It is possible to realize a natural spoken dialogue simulating the tendency of a real dialogue in which the prosody of a spoken voice changes over time. In the fourth embodiment, in particular, the standby time W, which is the interval between the uttered voice Vx and the response voice Vy, is controlled according to the usage history H. Therefore, at the stage immediately after starting the dialogue in the first meeting, the interval between the utterance and the response is long, and the interval is shortened as the dialogue with the dialogue partner is repeated. Voice dialogue is realized.

<Modification>
The voice interaction device 100 (100A, 100B, 100C, 100D) exemplified in each of the above embodiments can be variously modified. Specific modifications will be described below. Two or more aspects arbitrarily selected from the following examples can be appropriately combined within a mutually consistent range.

(1) It is also possible to combine any two or more configurations selected from the first to fourth embodiments. Specifically, the configuration of the first embodiment in which the prosody of the response voice Vy is controlled in accordance with the prosody (for example, pitch P) of the utterance voice Vx is similarly applied to the second to fourth embodiments. obtain. For example, in the second embodiment, after the prosody of the response signal Y selected in step SB3 or step SB4 in FIG. 9 is controlled according to the prosody (for example, pitch P) of the uttered voice Vx, the reproduction is performed by the reproducing device 26. Is also possible. Similarly, in the third embodiment, a configuration is adopted in which the prosody of the response signal Y selected in step SC2 or SC3 in FIG. 12 is controlled in accordance with the prosody of the speech voice Vx. In the fourth embodiment, the configuration shown in FIG. A configuration may be adopted in which the prosody of the response signal Y obtained from the storage device 22 in step SD5 is controlled in accordance with the prosody of the uttered voice Vx. In the configuration in which the first embodiment is applied to the second to fourth embodiments, for example, at the start point of a specific mora (typically, the last mora) in the response voice Vy, as in the first embodiment. The pitch of the response signal Y is adjusted such that the pitch matches the lowest pitch Pmin in the last section E of the uttered voice Vx.

  The configuration of the third embodiment in which the first response voice Vy1 of the interrogation for the utterance voice Vx and the second response voice Vy2 other than the interrogation are selectively reproduced may be applied to each mode other than the third embodiment. It is possible. Further, the configuration of the fourth embodiment that controls the prosody (for example, the standby time W) of the response voice Vy according to the usage history H of the voice dialogue can be applied to the first to third embodiments. .

(2) Various variables related to the above-described various forms of voice dialogue are variably set in accordance with, for example, an instruction from the user U. For example, a configuration in which the reproduction volume of the response voice Vy is controlled in accordance with an instruction from the user U, or the actual reproduction of a plurality of types of response voices Vy having different sexes or voice qualities (gentle voice, severe voice). A configuration in which the type of the response voice Vy reproduced from the device 26 is selected according to an instruction from the user U may be adopted. In the first to third embodiments, the length of the standby time W from the end point tB of the utterance voice Vx to the response reproduction point ty of the response voice Vy may be set in accordance with an instruction from the user U. It is possible.

(3) In the modified example of the third embodiment, the reproduction frequency of the first response voice Vy1 of the inquiry to the utterance voice Vx is variably set according to the instruction from the user U. It is also possible to control the reproduction frequency of the first response voice Vy1 according to the factor of. Specifically, a configuration in which the response generation unit 36D of the third embodiment controls the reproduction frequency of the first response voice Vy1 in accordance with the usage history H exemplified in the fourth embodiment may be employed. For example, in a dialogue between real humans, as the dialogue with a specific dialogue partner is repeated, the characteristics of the utterance (eg, habit and tone) of the dialogue partner can be grasped, and as a result, the frequency of querying the uttered voice decreases. Is assumed. In consideration of the above tendency, it is preferable that the frequency of reproduction of the first response voice Vy1 be reduced as the number of uses N indicated by the use history H increases.

(4) In the fourth embodiment, the use frequency N of the voice interaction is exemplified as the use history H, but the use history H is not limited to the use frequency N. For example, the number of times the response voice Vy is reproduced once in the voice interaction, the frequency of use of the voice interaction (the number of uses per unit time), the usage period of the voice interaction (for example, from the first use of the voice interaction device 100). It is also possible to apply the elapsed time since the last use of the voice interactive device 100 to the control of the standby time W as the usage history H.

(5) In the first embodiment, the response signal Y is generated and reproduced from the audio signal Z stored in the storage device 22 in advance. In the second to fourth embodiments, the response signal Y is stored in the storage device 22 in advance. Although the response signal Y is reproduced, the response signal Y indicating the response voice Vy of the specific utterance content can be synthesized by, for example, a known voice synthesis technique. For the synthesis of the response signal Y, for example, speech synthesis of a unit connection type or speech synthesis using a statistical model such as a hidden Markov model is suitably used. Further, the uttered voice Vx and the response voice Vy are not limited to human uttered sounds. For example, the animal's cry can be used as the utterance voice Vx or the response voice Vy.

(6) In each of the above-described embodiments, the configuration in which the voice interaction device 100 includes the voice input device 24 and the playback device 26 has been described as an example. It is also possible to provide an input device 24 and a playback device 26. The voice interaction device 100 is realized by a terminal device such as a mobile phone or a smartphone, and the voice input / output device is realized by an electronic device such as an animal-type toy or a robot. The voice interaction device 100 and the voice input / output device can communicate wirelessly or by wire. That is, the utterance signal X generated by the voice input device 24 of the voice input / output device is transmitted to the voice interactive device 100 wirelessly or wired, and the response signal Y generated by the voice interactive device 100 is transmitted wirelessly or wiredly to the voice input / output device. It is transmitted to the playback device 26.

(7) In each of the embodiments described above, the voice interactive device 100 is realized by an information processing device such as a mobile phone or a personal computer. However, some or all of the functions of the voice interactive device 100 are implemented by a server device (a so-called cloud server). It is also possible to realize. Specifically, the voice interaction device 100 is realized by a server device that communicates with a terminal device via a communication network such as a mobile communication network or the Internet. For example, the voice interaction device 100 receives the speech signal X generated by the voice input device 24 of the terminal device from the terminal device, and generates a response signal Y from the speech signal X by the configuration according to each of the above-described embodiments. Then, the voice interaction device 100 transmits the response signal Y generated from the speech signal X to the terminal device, and causes the reproducing device 26 of the terminal device to reproduce the response voice Vy. The voice interaction device 100 is realized by a single device or a group of a plurality of devices (that is, a server system). In addition, some functions of the voice interaction device 100 according to each of the above-described embodiments (for example, at least a part of the voice acquisition unit 32, the voice analysis units 34A and 34C, the response generation units 36A, 36B, 36C and 36D, and the history management unit 38) ) Can be realized by a server device, and other functions can be realized by a terminal device. Whether each function realized by the voice interaction device 100 is realized by the server device or the terminal device (function sharing) is arbitrary.

(8) In each of the above-described embodiments, the response voice Vy of a specific utterance content (for example, Aizuchi such as “un”) is reproduced with respect to the utterance voice Vx, but the utterance content of the response voice Vy is limited to the above examples. Not done. For example, the utterance content of the utterance voice Vx is analyzed by voice recognition and morphological analysis of the utterance signal X, and a response voice Vy having appropriate content for the utterance content is selected from a plurality of candidates and reproduced by the reproduction device 26. Is also possible. In a configuration in which speech recognition or morphological analysis is not performed (for example, the first to fourth embodiments), a response voice Vy of the utterance content prepared in advance is reproduced regardless of the utterance voice Vx. Therefore, when simply considered, it can be inferred that a natural dialogue is not established. However, as the prosody of the response voice Vy is variously controlled as illustrated in the above-described respective embodiments, actually, humans can not communicate with each other. The user U can perceive a feeling like a natural conversation. On the other hand, according to the configuration in which the speech recognition and the morphological analysis are not performed, there is an advantage that the processing delay and the processing load caused by these processes are reduced or eliminated.

(9) The voice interaction device 100 (100A, 100B, 100C, 100D) exemplified in each of the above-described embodiments can be used for evaluating an actual dialogue between humans. For example, when the prosody of a response voice (hereinafter referred to as “observation voice”) observed in an actual dialogue between humans is compared with the prosody of the response voice Vy generated in the above-described form, and the prosody is similar between the two. , It is possible to evaluate the observed speech as inappropriate while the prosodic divergence between the two is appropriate. The device (dialogue evaluation device) that executes the evaluation exemplified above can also be used for training of human interaction.

(10) The voice interaction device 100 (100A, 100B, 100C, 100D) exemplified in each of the above-described embodiments can be realized by cooperation between the control device 20 and the voice interaction program as described above. The voice dialogue program may be provided in a form stored on a computer-readable recording medium and installed on the computer. The recording medium is, for example, a non-transitory recording medium, and an optical recording medium (optical disk) such as a CD-ROM is a good example, and a known arbitrary recording medium such as a semiconductor recording medium or a magnetic recording medium is used. In the form of a recording medium. It is also possible to distribute the program to a computer in the form of distribution via a communication network. Further, the present invention can also be realized as an operation method (voice interaction method) of the voice interaction device 100 exemplified in each of the above-described embodiments. The computer (speech dialogue device 100) that is the main subject of the speech dialogue method is, for example, a system composed of a single computer or multiple computers.

100 (100A, 100B, 100C, 100D) ... voice interaction device, 20 ... control device, 22 ... storage device, 24 ... voice input device, 242 ... sound collection device, 244 ... A / D converter , 26 playback device, 262 D / A converter, 264 sound emitting device, 32 voice acquisition unit, 34A, 34C voice analysis unit, 36A, 36B, 36C, 36D response generation Unit, 38: History management unit.

Claims (4)

  A method for performing a voice dialogue that plays a response voice to an uttered voice,
  In the past, the playback device reproduces a prosody response voice according to the number of voice conversations, the frequency of use of voice conversations, the use period of voice conversations, or the elapsed time since the last voice conversations.
  A spoken dialogue method implemented by a computer.   The prosody of the response voice is a standby time that is an interval between the utterance voice and the response voice.
  The voice interaction method according to claim 1. An apparatus for executing a voice dialogue for reproducing a response voice to an uttered voice,
A voice having a response generation unit for causing a playback device to reproduce a prosody response voice corresponding to the number of voice conversations, the frequency of voice conversations, the use period of voice conversations, or the elapsed time since the last voice conversations in the past. Interactive device. A program for voice dialogue for reproducing a response voice to an utterance voice,
Computer
A program that functions as a response generation unit that causes a playback device to reproduce a prosodic response voice according to the number of voice dialogues, the frequency of voice dialogue use, the duration of voice dialogue, or the elapsed time since the last voice dialogue in the past. .

Images