JP2009053581A - Speech output device - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To obtain a speech output device changing a speaking pace to a comfortable one by a simple method. <P>SOLUTION: The speech output device which changes a speaking speed of output speech or a pause length, or both of them, by inputting speech, comprises: a speech detection section 110 for detecting input of speech; a speech output section 150 for outputting predetermined output speech; and a control section 120 for controlling the speaking speed of output speech or the pause length, or both of them. The control section 120 controls the speaking speed of output speech or the pause length, or both of them, of a following block, based on elapsed time from when the speech output section 150 outputs the output speech of one block, to when the speech detection section 110 detects the input of speech. <P>COPYRIGHT: (C)2009,JPO&INPIT

Description

  The present invention relates to an audio output device that can vary the speech speed and / or pause length of output audio.

  In an apparatus that outputs a predetermined voice and provides a voice guidance to a user, such as a voice guidance system, a plurality of voice blocks (words, phrases, sentences, paragraphs, etc.) are usually output continuously. There has been a problem that the pace at which the voice is output (hereinafter referred to as the utterance pace) does not match the utterance pace desired by the user, and voice that is difficult to hear for the user is presented.

  Therefore, with regard to the speech synthesizer, “in relation to a text-to-speech synthesizer that reads out a text composed of a plurality of sentences in units of sentences, the user controls the timing to start reading a sentence in the middle and understands the contents. It is possible to move to speech synthesis. As a technology for the purpose of the above, “a delimiter information identifying unit 3 that identifies predetermined data included in or added to text data as delimiter information, and a voice in response to an identification signal from the delimiter information identifying unit 3 Responsive to a restart signal from the start information identifying unit 8, a synthesis interruption control unit 4 that instructs to interrupt the synthesis operation, a start information identifying unit 8 that identifies predetermined voice information based on the user's utterance as output start information And a synthesis start control unit 9 for instructing the start of the speech synthesis operation corresponding to the next sentence of the text data interrupted. Is proposed (Patent Document 1).

JP-A-8-248990 (Summary)

The speech synthesizer described in Patent Document 1 has a pause / resume function and a function for detecting a user's utterance, and waits for a user's confirmation voice for each voice presentation block (for example, “one sentence”). Thus, the next audio block can be reproduced.
According to this voice synthesizer, the user can listen to the voice at his / her own pace while checking the voice content for each block.

  However, the technique described in Patent Document 1 mainly focuses on prevention of overhearing and the like, and there is a problem with ensuring user comfort when listening to audio. For example, there are the following problems (1) to (2).

(1) Since it is necessary to make a confirmation utterance for each voice block, a burden is imposed on the user. For example, when listening to sound such as news (one sentence for one block), it is necessary to make a confirmation utterance or confirmation for each sentence, which is troublesome.
(2) Although indirect utterance control by confirmation utterance can be performed, the utterance pace itself cannot be changed to a comfortable one for the user.

  Therefore, there has been a demand for an audio output device that can vary the utterance pace to be comfortable for the user by a simple method.

  An audio output device according to the present invention is an audio output device that changes a speech speed and / or pause length of an output audio by inputting the audio, and includes an audio detection unit that detects the input of the audio, An audio output unit that outputs an output audio; and a control unit that controls a speech speed and / or pause length of the output audio, and the control unit outputs the output audio of one block. After that, the speech speed and / or pause length of the output voice of the next block is controlled based on the elapsed time from when the voice detection unit detects voice input.

  According to the audio output device of the present invention, the speech speed and pause length of the output sound of the next block are controlled based on the elapsed time from the output of the audio block until the audio input is detected. If it is desired to speed up, the speech pace can be changed to a comfortable one for the user by a simple method such as inputting a confirmation voice immediately after the end of the voice block.

Embodiment 1 FIG.
FIG. 1 is a functional block diagram of an audio output device 100 according to Embodiment 1 of the present invention.
The audio output device 100 is a device that outputs predetermined output audio, and can change the speech speed and pause length according to the timing when the user inputs confirmation audio, and adjust these so that the user can easily hear them. It is. The “pause” here refers to a silent section between speech blocks.
Hereinafter, the configuration of the audio output device 100 will be described.

  The voice output device 100 includes a voice detection unit 110, an utterance pace control unit 120, a variable rule table 130, a voice database (hereinafter referred to as a voice DB) 140, and a voice output unit 150.

The voice detection unit 110 receives any voice data input by the user as an input, and outputs a voice input start notification to the utterance pace control unit 120.
Specifically, after an audio signal obtained via a microphone is AD converted, an audio detection process is performed by digital signal processing. Any detection method may be used as long as it is a general method used for voice detection. For example, the power of a voice signal is calculated in units of frames (eg, length 20 ms, period 5 ms) and has a power of a certain level or more. If the frame continues for a predetermined time (e.g., 3 to 5 frames), it is determined that the voice section has started.
In addition, detection accuracy may be further increased by creating a detection determination rule from the maximum value of the autocorrelation function, the mel cepstrum, and its difference.
When the voice detection unit 110 determines that the voice section has started, it immediately outputs a voice input start notification.

  Based on the voice input start notification received from the voice detection unit 110 and the voice output end notification and voice output start notification received from the voice output unit 150, the utterance pace control unit 120 allows the voice output unit 150 to change the utterance pace. Direct maintenance.

  The speech pace control unit 120 includes a timing analysis unit 121 and a voice output instruction unit 122 as internal configurations.

  The timing analysis unit 121 outputs three types of timing analysis results (UP / KEEP / DOWN) to the audio output instruction unit 122 based on the reception timing of each notification described in FIG.

The audio output instruction unit 122 receives an audio output end notification and an audio output start notification from the audio output unit 150, and receives a timing analysis result from the timing analysis unit 121. Furthermore, a speech speed selection instruction and a pause length adjustment instruction are output to the voice output unit 150.
The contents of the speech speed selection instruction and pause length adjustment instruction are as follows (1) to (3). Details thereof will be described later with reference to FIG.

(1) When “UP” is received from the timing analysis unit 121, the speech speed is increased, the pause length is decreased, or both are performed.
(2) When “KEEP” is received from the timing analysis unit 121, the speech speed and / or pause length are maintained.
(3) When “DOWN” is received from the timing analysis unit 121, the speech speed is slowed down, the pause length is lengthened, or both are performed.

  The variable rule table 130 stores variable rules described later with reference to FIG.

The audio database 140 stores audio data (for example, a wav file, the same applies hereinafter) corresponding to each output audio for each audio block.
Note that the voice database 140 includes voice data corresponding to a plurality of speech speeds for the output voice having the same content. For example, voice data uttered at a plurality of speech speeds (such as fast / normal / slow) is stored for each output voice having the same content.

The voice output unit 150 receives a speech speed selection instruction and a pause length adjustment instruction from the voice output instruction unit 122, reads voice data of an appropriate speech speed from the voice DB 140 based on the instruction, and outputs the voice with an appropriate pause length.
When the voice output is started, a voice output start notification is output to the timing analysis unit 121 and the voice output instruction unit 122 when the voice output is ended.

The utterance pace control unit 120 can be configured by hardware such as a circuit device that realizes the function, or can be configured as software executed on an arithmetic device such as a microcomputer or CPU.
The variable rule table 130 and the voice database 140 can be configured by data files necessary for configuring them, and a storage device such as a memory or HDD (Hard Disk Drive) for storing the data files. An appropriate file format may be used as appropriate.

Note that the “control unit” in the first embodiment corresponds to the speech pace control unit 120.
The “storage unit” corresponds to a storage device that configures the variable rule table 130.

Heretofore, each configuration of the audio output device 100 has been described.
Next, the timing analysis contents of the timing analysis unit 121 will be described with reference to FIG. 2, but prior to this, the basic concept of timing analysis will be described.

In general conversations, if you want the other party to speak more quickly, you often prompt the other person to speak immediately after the other person's utterance has been completed by hitting the conversation immediately. On the other hand, when you want the other party to speak more slowly, the opposite is often the case.
Therefore, this operation is also incorporated in the audio output device 100 according to the present invention.
That is, when the voice output device 100 finishes outputting one block of voice and the confirmation voice is input immediately, it is determined that the user feels “I want you to speak sooner” and the confirmation voice is If it takes time to input, it is determined that the user feels "I want to speak more slowly".

When the timing analysis unit 121 determines that the user feels "I want you to speak sooner", outputs "UP", and when the user determines that the user feels "I want you to speak more slowly" Outputs “DOWN”. If it is between these, “KEEP” is output.
An example of these criteria will be described with reference to FIG.

FIG. 2 is an example of determination criteria for the timing analysis unit 121 to vary the speech speed and pause length based on the timing at which a voice input start notification is received. In the following, description will be made separately for each timing of receiving a voice input start notification.
Note that the thresholds TH1 and TH2 shown in FIG. 2 are in a relationship of TH1 <TH2.

(A) When the audio output end notification is received at TH1 The timing analysis unit 121 receives an audio output end notification from the audio output unit 150, and from the audio detection unit 110 until the threshold TH1 elapses. When the voice input start notification is received, it is determined that the user feels “I want to speak sooner”, and “UP” is output.

(B) When received by TH1 to TH2 The timing analysis unit 121 receives an end notification of one block of audio output from the audio output unit 150, and after the threshold value TH1 has elapsed, the threshold value TH2 has elapsed. When a voice input start notification is received from the detection unit 110, “KEEP” is output to maintain the current speech speed and pause length.

(C) When received after TH2 and before the next audio output start notification The audio input start notification from the audio detection unit 110 until the timing analysis unit 121 receives the next audio output start notification after the threshold TH2 has elapsed. Is received, it is determined that the user feels “I want to speak more slowly”, and “DOWN” is output.

(D) When received by voice output start notification to TH3 The timing analysis unit 121 receives the voice output start notification of one block from the voice output unit 150, and from the voice detection unit 110 until the threshold TH3 has elapsed. When the voice input start notification is received, it cannot be determined whether the user's response to the previous voice block is delayed or whether the current voice block is immediately responded.
Therefore, in order to maintain the current speech speed and pause length, “KEEP” is output.

(E) When TH3 is received as an audio output end notification The timing analysis unit 121 receives an audio output start notification from the audio output unit 150, and after the threshold TH3 has elapsed, the audio block from the audio detection unit 110 If the voice input start notification is received from the voice detection unit 110 until the voice output end notification is received, it is determined that the example is the same as (a), and “UP” is output.

Note that the values of the thresholds TH1, TH2, and TH3 are determined in advance by analyzing actual dialogue data and the like based on the viewpoint of the concept that is the basis of the timing analysis described above. For example, the number of sound output mora (beats) is determined as TH1 = 1 mora, TH2 = 3 mora, TH3 = 1 mora, and the like.
However, it is preferable to change the values of TH1 to TH3 depending on the speech speed and pause length. Therefore, it is preferable to determine for each block according to the speech speed and pause length of the speech block.
In the first embodiment, TH1 to TH3 are determined according to the speech speed, and are stored in a threshold table (not shown) in the timing analysis unit 121.

FIG. 3 shows the configuration of the variable rule table 130 and data examples.
The variable rule table 130 stores rules for the voice output instruction unit 122 to instruct to change the speech speed and pause length based on the result of the timing analysis performed by the timing analysis unit 121.
In the data example of FIG. 3, the initial values of the pause length and speech speed are P0 and V0, respectively.
Upon receiving “UP”, “KEEP”, and “DOWN” from the timing analysis unit 121, the voice output instructing unit 122 instructs to change the speech speed and pause length based on the rules shown in the variable rule table 130 as follows. I do.

(1) When UP is received As the speech speed selection instruction, +1, which is one stage “faster” than the current speech speed, is output to the voice output unit 150.
As a pause length adjustment instruction, +1, which is “one step shorter than the current pause length”, is output to the audio output unit 150.
(2) When KEEP is received, nothing is output.
(3) When DOWN is received As the speech speed selection instruction, “-1” that is “slower” than the current speech speed is output to the voice output unit 150.
As a pause length adjustment instruction, “-1” that is “longer than” the current pause length is output to the audio output unit 150.

FIG. 4 shows an operation example of the audio output device 100.
Here, it is assumed that the audio output device 100 is a device in which a microphone and a speaker are connected, and the user sits in front of the microphone and listens to the audio.
The voice output device 100 outputs a voice message (for example, “Today's news”) from the speaker for each sentence.
The user listens to this voice message (while whispering) and occasionally speaks out. For example, “Yes”, “Hey”, “I see”, “Yes,” “Next!”, “Slightly wait”.
The sound emitted from the user is taken into the sound output device 100 from the microphone, and sound detection is performed.
The operation of the audio output device 100 will be described below step by step according to FIG.

(1) The timing analysis unit 121 sets pause length = P0 and speech speed = V0 as initial values.
(2) The audio output unit 150 starts audio output of the first audio block (B1). As the voice data at this time, data that matches the speech speed initial value V0 in the voice DB 140 is used.
(3) If no confirmation voice input from the user is detected after the audio block B1 ends and before the pause length initial value P0 elapses, the audio is transmitted when the pause length P0 elapses after the audio block B1 ends. The output unit 150 immediately starts outputting the sound of the second sound block (B2).

(4) Assume that the user utters the voice S2 after the end of the voice block B2 and before the pause length P0 has elapsed.
(5) Upon detecting the voice S2, the voice detection unit 110 outputs a voice input start notification to the timing analysis unit 121 immediately. This output time is slightly delayed from the actual speech start time because the voice detection unit 110 performs frame processing, but the delay is ignored here.

(6) The timing analysis unit 121 determines (UP / KEEP / DOWN) based on the rule described above with reference to FIG. 2 based on the reception time of the voice input start notification and the reception time of the voice block B2 end notification. To do. The determination result is immediately output to the voice output instruction unit 122.
In the example of FIG. 4, since the voice input start notification is received during the period from TH1 to TH2, “KEEP” is output.

(7) Since the voice output instruction unit 122 has received the “KEEP” message, the voice output instruction unit 122 does not issue an instruction to change the speech speed or pause length, and maintains the previous setting as it is.
(8) Since the voice output unit 150 does not receive an instruction from the voice output instruction unit 122 during the pause period, when the pause length P0 has elapsed after the end of the voice block B2, the third voice block (B3 ) Audio output.

(9) Assume that the user utters the voice S3 after the end of the voice block B3 and before the pause length P0 elapses.
(10) The voice detection unit 110 outputs a voice input start notification to the timing analysis unit 121 as soon as the voice S3 is detected.
(11) The timing analysis unit 121 determines (UP / KEEP / DOWN) as in step (6), and outputs the determination to the voice output instruction unit 122. Here, since the elapsed time from the reception of the voice input start notification is less than TH1, “UP” is output.

(12) Since the voice output instruction unit 122 has received the “UP” message, the voice output instruction unit 122 instructs the voice output unit 150 to change the speech speed V to +1 and the pause length P to +1.
(13) The voice output unit 150 changes the pause length to P0 + 1 and the speech speed to V0 + 1, and writes them in the internal pause length memory and the speech speed memory (both not shown).
(14) The audio output unit 150 reads the audio data of the fourth audio block (B4) corresponding to the speech speed V0 + 1 from the audio DB 140, and after the audio block B3 ends and after the pause length P0 + 1 has elapsed, the audio of the audio block B4 Start output.

(15) It is assumed that the user utters the voice S4 before the pause length P0 + 1 elapses after the voice block B4 ends.
(16) The voice detection unit 110 outputs a voice input start notification to the timing analysis unit 121 as soon as the voice S4 is detected.
(17) The timing analysis unit 121 determines (UP / KEEP / DOWN) as in step (6), and outputs the determination to the voice output instruction unit 122. Here, since the elapsed time from the reception of the voice input start notification is TH2 or more, “DOWN” is output.

(18) Since the voice output instruction unit 122 has received the “DOWN” message, the voice output instruction unit 122 instructs the voice output unit 150 to change the speech speed V to −1 and the pause length P to −1.
(19) The voice output unit 150 changes the pause length to P0 and the speech speed to V0, and writes them in the internal pause length memory and the speech speed memory (both not shown).
(20) The audio output unit 150 reads the audio data of the fifth audio block (B5) corresponding to the speech speed V0 from the audio DB 140, and after the audio block B4 ends and after the pause length P0 has elapsed, the audio of the audio block B5 Start output.

  As described above, according to the first embodiment, the utterance pace control unit 120 has elapsed time from when the voice output unit 150 outputs a voice block to voice, until the voice detection unit 110 detects the user's voice input. Since the speech speed and pause length of the output voice are controlled based on the above, it is possible to adjust so that the voice output is performed at the utterance pace matching the user's pace only by the user having a simple reconciliation.

Moreover, since the speech pace control unit 120 determines the thresholds TH1 to TH3 for each block according to the speech speed and pause length of each speech block, an optimal threshold is set according to the speech pace of each speech block. It is possible to respond to the user in detail.
For example, when the pause length is set to be long, there is sufficient time for the user to input the confirmation voice. Therefore, it is considered that the thresholds TH1 to TH3 should be set longer, as described above. By setting these threshold values in accordance with the speech pace of each voice block, it is possible to set the threshold in accordance with the substance of the voice block.

Embodiment 2. FIG.
In the first embodiment, which voice data stored in the voice DB 140 is used or what the pause length is to be used based on the pause length and speech speed variable rules held in the variable rule table 130. Explained that to determine.
In the second embodiment of the present invention, it will be described that by changing the contents of the variable rule table 130, the variable operation of speech speed and pause length is different from that of the first embodiment.
Since other configurations are the same as those of the first embodiment, description thereof is omitted.

FIG. 5 shows another configuration example of the variable rule table 130.
In FIG. 3 described in the first embodiment, the example of the variable rule table in which the pause length P and the speech speed V increase or decrease in conjunction with the determination of “UP” and “DOWN” has been described. The setting is not limited and can be set according to the usage environment of the audio output device 100 and the contents of the audio DB 140.

FIG. 5A is an example of a rule in which priority is given to the adjustment of the pause length, and when the pause length adjustment is saturated, the speech speed is adjusted.
FIG. 5B shows a rule that prioritizes the adjustment of the speech speed, contrary to FIG. 5A.
FIG. 5C shows a rule in which adjustment of speech speed and pause length is mixed.

  These various rules can be changed by replacing the variable rule data stored in the variable rule table 130.

  In this way, by configuring the variable rule table 130 so as to be changeable, it is not necessary to create each configuration one by one according to the individual usage status of the audio output device 100, and it is only necessary to replace the data file. Therefore, the flexibility of the configuration is increased, and the audio output device 100 can be easily adapted to more applications and environments.

Embodiment 3 FIG.
In the first and second embodiments, the timing analysis unit 121 performs the determination described with reference to FIG. 2 based on the reception time of the voice input start notification (and the voice output end notification). Occurs.
Therefore, in Embodiment 3 of the present invention, a method for reducing the delay due to such frame processing will be described.

When the voice detection unit 110 outputs a voice input start notification to the timing analysis unit 121, the speech start time itself is inserted in the notification message. Similarly, when the sound output unit 150 outputs a sound output end notification to the timing analysis unit 121, the sound output end time itself is inserted.
The timing analysis unit 121 obtains these times included in the notification instead of the reception times of these notifications, and performs the determination described with reference to FIG. 2 based on the values.
As described above, the delay due to the frame processing can be reduced by making the determination based on the time information included in the notification instead of the reception time of the notification.

  FIG. 6 illustrates a procedure in which the timing analysis unit 121 acquires the utterance start time of the confirmation voice of the user. The outline of each step in the figure will be described below.

(1) When the user starts inputting confirmation voice, detection of the voice signal by the voice detector 110 is started.

(2) The voice detection unit 110 does not immediately output a voice input start notification when the user starts input of the confirmation voice, but, for example, has a power of a certain level or more as described in the first embodiment. If the frame possessed continues for a predetermined time (for example, 3 to 5 frames), it is determined that the voice section has started. Therefore, the voice detection unit 110 performs frame accumulation of the voice signal.
While accumulating frames, the power value calculation as described above is performed in parallel.

(3) When the above step (2) is executed and it is determined that the voice section is started, the voice detection of the confirmation voice is completed.
(4) When the voice detection of the confirmation voice is completed, the voice detection unit 110 outputs a voice input start notification to the timing analysis unit 121 immediately (but with a time lag).
(5) The input of the user confirmation voice is completed.

  As shown in FIG. 6, there is a time lag associated with the frame processing between when the user starts inputting the confirmation voice and when the timing analysis unit 121 receives the voice input start notification. Therefore, in anticipation of this time lag, a more accurate utterance start time can be acquired by the following method.

(Method 1) The time of voice detection completion is included.
When the voice detection unit 110 outputs a voice input start notification to the timing analysis unit 121, the voice detection unit 110 includes the time when the voice detection is completed as the utterance start time.
Even if there is a time lag between when the user starts input of the confirmation voice and when the timing analysis unit 121 receives the voice input start notification, by referring to the voice detection completion time included in the notification, It approaches the actual voice input start time.

(Method 2) Allow for frame accumulation time required for voice detection.
In step (2) of FIG. 6, the voice detection unit 110 anticipates the frame accumulation time necessary for determining that “the voice section has started”, and further detects this frame from the voice detection completion time described in the method 1 above. Subtract the accumulation time. The voice detection unit 110 includes the time after the subtraction in the voice input start notification as the utterance start time.
Thereby, it can be brought closer to the actual voice input start time.

(Method 3) The time required for the calculation is subtracted.
In addition to the above method 2, the voice detection unit 110 subtracts the calculation time required for calculating the power of the voice signal and the time after the subtraction is included in the voice input start notification as the utterance start time. Since the calculation time seems to be short compared with other times, this method may be omitted.

(Method 4) Back-calculate from sampling frequency and sampling sequence number.
In the above methods 1 to 3, when the confirmation voice is digitally recorded, the time can be calculated backward from the sampling frequency and the sampling sequence number.
In this case, the voice detection unit 110 samples and stores the user's confirmation voice as digital voice data when storing the frame. A serial number is assigned to each sample.
For example, in method 2, when subtracting the frame accumulation time, instead of subtracting the time itself, the time to be subtracted is calculated by multiplying the number up to the return sample number by the sampling frequency. Can do.

  In the above (Method 1) to (Method 4), the voice input start notification has been described. However, for each of the other notifications, an accurate time can be acquired by adopting the same method.

Even when any of the above (Method 1) to (Method 4) is used, the timing analysis unit 121 acquires the utterance start time included in the voice input start notification, so that the user has actually started to input the confirmation voice. The time as close as possible to the time can be acquired.
Thereby, since the determination based on the threshold values TH1 to TH3 described in FIGS. 2 to 4 is more accurate, the possibility that the threshold determination that is not intended by the user is performed is reduced, which contributes to the convenience of the user.

  Although it has been described that the time after subtraction is included in each notification, the subtraction process may be performed after each notification is received.

Embodiment 4 FIG.
FIG. 7 is a functional block diagram of the audio output device 100 according to Embodiment 4 of the present invention.
The voice output device 100 according to the fourth embodiment includes an utterance text 160 and a voice synthesizer 170 instead of the voice DB 140 described in FIG. 1 of the first embodiment. Since other configurations are the same as those of the first to third embodiments, the description thereof is omitted.

In the first embodiment, voice data (wav file) corresponding to the utterance content is stored in the voice DB 140 for each speech speed in advance, and this is read out and output as voice.
In the fourth embodiment, only the text of the utterance content is stored as the utterance text 160, and the output speech is dynamically generated by speech synthesis. When changing the speech speed, the speech speed is given as a designated parameter during speech synthesis.

The utterance text 160 can be configured by storing a data file such as a text file in a storage device such as a memory or HDD. The utterance text 160 is divided for each voice block by a delimiter or a file division.
The speech synthesizer 170 performs speech synthesis based on the content of the utterance text 160, and a technique generally known as TTS (Text To Speech) can be used. Note that the above-described speech speed designation parameter is received as an external parameter, and speech synthesis is performed so that the output speech has a speech speed according to the parameter.

Next, the operation of the audio output device 100 according to the fourth embodiment will be described.
The operation of the audio output device 100 according to the fourth embodiment is substantially the same as that described in the first to third embodiments. The following is a brief description focusing on the differences.

The voice output unit 150 requests the voice synthesizer 170 to perform voice synthesis at the speaking speed instructed from the voice output instruction unit 122.
The speech synthesizer 170 reads text data corresponding to one sentence from the utterance text 160, synthesizes speech at a normal speech speed, and outputs the speech from the speech output unit 150.
Then, when the voice detection unit 110 detects the input of the confirmation voice from the user during the pause period, the voice output instruction unit 122 instructs the voice output unit 150 on the speech speed and pause length based on the analysis result of the timing analysis unit 121. To do.
The speech synthesizer 170 synthesizes speech at the speaking speed instructed by the speech output unit 150 and outputs it to the speech output unit 150. The voice output unit 150 outputs the synthesized voice received from the voice synthesis unit 170 as a voice.

  As described above, according to the fourth embodiment, since only one type of utterance text 160 is prepared, voice output at an arbitrary speaking speed can be dynamically performed by voice synthesis. It is no longer necessary to prepare audio data (wav file) corresponding to the speed, and the time and effort of the previous construction is reduced.

  In addition, although the audio data generally has a huge data size, according to the configuration of the fourth embodiment, only the text data needs to be stored, so the data size can be small, and the audio output device 100 can be small. Contributes to cost reduction and cost reduction.

Embodiment 5 FIG.
In the first to fourth embodiments, the user inputs an arbitrary confirmation voice, and the timing analysis unit 121 analyzes the timing at which the voice detection unit 110 detects the confirmation voice according to the determination criteria as shown in FIG. And explained how to change the pose length.
In the fifth embodiment of the present invention, a configuration is described in which the confirmation voice input by the user is limited to a specific word or phrase.

FIG. 8 is a functional block diagram of the audio output device 100 according to the fifth embodiment.
In FIG. 8, a voice recognition unit 180 is newly provided between the voice detection unit 110 and the utterance pace control unit 120. Other configurations are the same as those in FIG. 1 described in the first embodiment, and thus description thereof is omitted. Note that the configurations of other embodiments may be used.

The voice recognition unit 180 receives the confirmation voice input by the user from the voice detection unit 110 and performs voice recognition processing. As a result of voice recognition, when it is determined that a predetermined reserved word has been input, a voice input start notification is output to the timing analysis unit 121.
The reserved word here is preferably one that clearly indicates that it is a confirmation voice, and may be a simple one such as “Yes” or “Yes”.

Hereinafter, the operation of the audio output device 100 according to the fifth embodiment will be briefly described.
When the user inputs a confirmation voice, the voice detection unit 110 detects the voice and outputs it to the voice recognition unit 180.
The speech recognition unit 180 performs speech recognition processing, outputs a speech input start notification to the timing analysis unit 121 if it is a predetermined reserved word, and outputs nothing if it is not a reserved word.
Since other operations are the same as those of the other embodiments, description thereof is omitted.

  As described above, according to the fifth embodiment, a sound other than the confirmation sound input by the user, such as a background sound, is not erroneously detected as the confirmation sound. Thereby, the possibility that the speech pace is changed unintentionally by the user is greatly reduced, which contributes to the convenience of the user.

Embodiment 6 FIG.
In Embodiments 1 to 5 described above, it has been described that the speech speed and pause length are controlled by the timing at which the user inputs the confirmation sound. However, instead of inputting the confirmation sound, the sound is input by another device such as a mouse. You can also.

Embodiment 7 FIG.
In the first embodiment, it has been described that voice data corresponding to each speech speed is stored in the voice DB 140. Instead, voice data corresponding to a single speech speed (for example, standard speech speed) is stored. It may be configured to store the voice and convert the voice speed by interposing a voice speed converter when outputting the voice.
Further, in the fourth embodiment, it has been described that the speech speed is instructed as an external parameter of the speech synthesizer 170, but instead, the speech speed is similarly converted by interposing a speech speed conversion device. May be.
Similarly, the other embodiments may be configured to convert the speech speed by interposing a speech speed conversion device.

Embodiment 8 FIG.
In the description of FIG. 2D of the first embodiment, it cannot be determined whether the user has responded late to the immediately preceding speech block or whether the user has responded immediately to the current speech block. “KEEP”.
In the eighth embodiment of the present invention, “WAIT” is defined as a more aggressive new operation, and this “WAIT” operation is performed for a user response in the same section.
Each configuration and basic operation are the same as those in the above-described embodiment, and thus description thereof is omitted.

In the eighth embodiment, when the timing analysis unit 121 receives the voice input start notification in the section of FIG. 2D, it outputs “WAIT”. This is similar to “KEEP” in that the speech speed and pause length are not changed, but is different from “KEEP” in that “WAIT” is internally stored.
Next, the significance of storing “WAIT” output internally and the specific operation will be described from the viewpoint of the user.

When the user detects the user's confirmation voice in the section of FIG. 2D, it is assumed that the user has input the confirmation voice in the section of FIG. 2C or FIG. Since the operation of the timing analysis unit 121 at this time is “WAIT”, the speech speed and the pause length do not change.
However, since the confirmation voice is input as the user intends to change the speech speed and pause length, the voice output device 100 does not operate as intended by the user.
In this case, when the user next inputs the confirmation voice, the input timing seems to be automatically finely adjusted so as to match the section of FIG. 2 (c) or (e).

As a result of the fine adjustment of the input timing, the next confirmation voice is performed in the section of FIG. 2 (c) or (e). If “UP” or “DOWN” is output for the first time at this time, the operation for one time is impaired from the viewpoint of the user, which impairs the user's feeling of use.
Therefore, using the fact that the information indicating that “WAIT” was output last time is stored internally, “UP” and “DOWN” are executed twice for the previous operation and the current operation. . Specifically, “UP” or “DOWN” may be output continuously twice, or the voice output instruction unit 122 may be instructed to double the control amount in one operation. .

  As described above, according to the eighth embodiment, “WAIT” is defined as a new operation, the fact that the previous “WAIT” is output is internally stored, and the operation is performed together with the previous operation at the next operation. Therefore, even when the input timing of the confirmation voice is as shown in FIG. 2D, a user-friendly voice output device can be provided without impairing the user's feeling after the next time.

Embodiment 9 FIG.
FIG. 9 is a configuration diagram of a voice guidance system according to Embodiment 9 of the present invention.
In FIG. 9, the audio output device 100 and the user terminal 200 are connected via a network 300.

The audio output device 100 has the configuration described in the first to eighth embodiments. It is assumed that an interface for connecting to the network 300 is appropriately provided.
The user terminal 200 is a computer that includes a microphone and a speaker.
The network 300 is a wired or wireless communication line.
Hereinafter, the operation of the voice guidance system of FIG. 9 will be briefly described.

(1) The user uses the user terminal 200 to request voice guidance from the voice output device 100 via the network 300.
(2) The voice output device 100 receives a request from the user terminal 200 and outputs a voice from the voice output unit 150. The output voice is transmitted as voice data to the user terminal 200 via the network 300.
(3) The user inputs confirmation voice into the microphone of the user terminal 200. The confirmation voice is transmitted as voice data to the voice output device 100 via the network 300.
(4) The voice detection unit 110 receives voice data of confirmation voice from the network 300. Subsequent operations are the same as those described in each embodiment.

  In the ninth embodiment, the user terminal 200 is a computer, but may be another terminal capable of voice input / output, for example, a mobile phone terminal. The network 300 is an appropriate communication network according to the type of terminal.

  As described above, according to the ninth embodiment, since the voice guidance by the voice output device 100 can be provided via the network 300, the voice output device 100 can be used in various usage forms.

3 is a functional block diagram of the audio output device 100 according to Embodiment 1. FIG. This is an example of a criterion for the timing analysis unit 121 to vary the speech speed and pause length based on the timing at which a voice input start notification is received. The structure of the variable rule table 130 and an example of data are shown. An example of the operation of the audio output device 100 is shown. 6 shows another configuration example of the variable rule table 130. The procedure by which the timing analysis unit 121 acquires the utterance start time of the confirmation voice of the user will be described. 6 is a functional block diagram of an audio output device 100 according to Embodiment 4. FIG. 10 is a functional block diagram of an audio output device 100 according to Embodiment 5. FIG. FIG. 20 is a configuration diagram of a voice guidance system according to a ninth embodiment.

Explanation of symbols

  DESCRIPTION OF SYMBOLS 100 Voice output device, 110 Voice detection part, 120 Speech pace control part, 130 Variable rule table, 140 Voice database, 150 Voice output part, 160 Speech text, 170 Voice synthesizer, 180 Voice recognition part, 200 User terminal, 300 Network .

Claims (8)

By inputting voice,
A voice output device that varies the speech speed and / or pause length of the output voice,
A voice detection unit for detecting voice input;
An audio output unit for outputting predetermined output audio;
A control unit for controlling the speech speed or pause length of the output voice or both;
With
The controller is
After the sound output unit outputs the output sound of one block,
Based on the elapsed time until the voice detection unit detects voice input,
An audio output device that controls the speech speed and / or pause length of the output audio of the next block. The controller is
If the elapsed time is less than a predetermined first threshold,
Increase the speech speed of the output sound of the next block, shorten the pause length, or both,
When the elapsed time is equal to or greater than a predetermined second threshold (where the first threshold is less than the second threshold)
Decrease the speaking speed of the output sound of the next block, increase the pause length, or both,
When the elapsed time is not less than the first threshold and less than the second threshold,
The speech output device according to claim 1, wherein the speech speed and / or pause length of the output speech of the next block is maintained. The controller is
When the voice detection unit detects a voice input after the voice output unit starts outputting the output voice and before a predetermined third threshold value elapses,
The speech output device according to claim 2, wherein the speech speed and / or pause length of the output speech of the next block is maintained. The controller is
When the voice detection unit detects voice input while the voice output unit starts outputting the output voice and outputs the output voice after the third threshold has elapsed,
The speech output apparatus according to claim 3, wherein the speech speed of the output speech of the next block is increased, the pause length is shortened, or both. The controller is
The first threshold, the second threshold, and the third threshold are:
The voice output device according to claim 3 or 4, wherein the voice output device is determined based on a speech speed and / or a pause length for each block of the output voice. A storage unit storing a variable rule of the speech speed or pause length of the output voice or both,
The audio output device according to claim 1, wherein the control unit controls the speech speed and / or pause length of the output audio with reference to the variable rule. The controller is
The time when voice input is started is received from the voice detector,
Alternatively, the time when the audio output is completed is received from the audio output unit,
The audio output device according to any one of claims 1 to 6, wherein the elapsed time is counted based on these times. The voice detection unit
Voice recognition means for recognizing the content of the input voice,
Only when the voice recognition means recognizes that a predetermined reserved word has been input,
The audio output device according to any one of claims 1 to 7, wherein it is determined that audio is input.

Images