Hotword detection on multiple devices

Generally, this specification relates to a system and a technique for recognizing a spoken word, which is also referred to as speech recognition.

In a real-life voice-operated home or other environment, that is, the user only needs to issue a query or command, and a computer-based system receives and answers the query and / or triggers the execution of the command. The environment exists. A voice-operated environment (eg, home, work, school, etc.) can be implemented by using a network of linked microphone devices distributed across various spaces or areas of the environment. Through such a microphone network, users can verbally contact the system from almost anywhere in the environment without having to have a computer or other device in front of or near them. For example, even while cooking in the kitchen, the user asks the system "how many milliliters in three cups?" And in response, the system outputs an answer, such as synthetic speech. Can be received in the form of. Alternatively, the user tells the system, "when does my nearest gas station. You can ask the question "close)" or the question "should I wear a coat today?" While preparing to go out.

In addition, users may query the system and / or issue commands regarding their personal information. For example, a user may ask the system "when is my meeting with John?" Or "remind me to call John when I get home." You can tell the system to call John when I get back home) ".

<p>The method by which the user interacts with the system for a system capable of voice operation is not limited to this, and is designed mainly by voice input. As a result, a system that potentially picks up all utterances that occur in the surrounding environment, including those that are not directed to the system, is one in which any given utterance is directed, for example, to someone present in the environment. Instead, we must have some means of identifying that it is aimed at the system. One way to achieve this is to use hotwords. Hotwords are reserved as predetermined words issued to draw the attention of the system by arrangement with users in the environment. In one example of the environment, the hotword used to draw the attention of the system is "OK Computer (OK). computer) ". Therefore, each time the word "OK computer" is spoken, it is picked up by the microphone and transmitted to the system. The system runs a speech recognition technique to determine if a hotword has been issued and, if so, waits for a subsequent command or query. Therefore, utterances directed at the system take the general form of [HOTWORD] [QUERY]. Here, "HOTWORD" in this example is "OK computer" and "QUERY" is any question that can be voice-recognized, analyzed, and executed by the system alone or in collaboration with the server over the network. It may be a command, notification, or other request.</p><p>According to an innovative aspect of the subject matter described herein, the user device receives utterances made by the user. The user device determines whether the utterance contains a hot word and calculates a hot word reliability score indicating that the utterance may contain a hot word. The user device sends this score to other nearby user devices. Perhaps other user devices are receiving the same utterance. Other user devices calculate hotword reliability scores and send those scores to the user device. The user device compares the hotword reliability scores. If the user device has the highest hotword reliability score, it will continue to operate and be prepared to process additional audio. The user device does not process additional audio if it does not have the highest hotword reliability score.</p><p>In general, another innovative aspect of the subject matter described herein may be implemented as a method, the method being the step of receiving audio data corresponding to an utterance by a first computing device and the utterance. Is the step of determining the first value corresponding to the possibility that the utterance contains the hot word, and the step of receiving the second value corresponding to the possibility that the utterance contains the hot word, and the second value is the second. Speech recognition processing for audio data based on the steps determined by the computing device, the step of comparing the first and second values, and the result of the comparison of the first and second values. Has a step to start with.</p><p>These and other embodiments may selectively include one or more of the following features, respectively. The method further comprises a step of determining that the first value has reached the hotword score threshold. The method further comprises a step of transmitting the first value to the second computing device. The method further comprises a step of determining the boot state of the first computing device based on the result of comparison between the first value and the second value. The step of determining the boot state of the first computing device based on the result of comparison between the first value and the second value further includes a step of determining that the boot state is the active state. The method is the step of receiving additional audio data corresponding to the additional utterance by the first computing device and the step of determining the third value corresponding to the possibility that the additional utterance contains a hot word. And the step of receiving a fourth value corresponding to the possibility that the additional utterance may contain a hotword, the fourth value being determined by the third computing device, the step and the third value. And a step of comparing the fourth value with, and a step of determining that the boot state of the first computing device is inactive based on the comparison result between the third value and the fourth value. Have.</p><p>The step of transmitting the first value to the second computing device further includes the step of transmitting the first value to the server via a local network or short-range wireless communication. The step of receiving a second value corresponding to the possibility that the utterance contains a hot word, the second value being determined by the second computing device, the step being determined by the second computing device. It further includes the step of receiving the second value from the server over the local network or over short range radio communication. The method further comprises a step of identifying the second computing device and a step of determining that the second computing device is configured to respond to an utterance containing a hotword. The step of transmitting the first value to the second computing device further includes the step of transmitting the first identifier to the first computing device. The step of receiving a second value corresponding to the possibility that the utterance contains a hot word, the second value is determined by the second computing device, the step is the second identifier for the second computing device. Includes further steps to receive. The step of determining that the activated state is the active state further includes a step of determining that a predetermined time has elapsed since the audio data corresponding to the utterance was received. The method further comprises a step of continuing to transmit the first value for a predetermined time based on the determination that the activation state is the active state.</p><p>Another embodiment of this embodiment comprises a computer program recorded on a corresponding system, device, and computer storage device, each configured to perform the operation of the above method.</p>

<p>Certain embodiments of the subject matter described herein may be implemented to realize one or more of the following advantages: Multiple devices can detect a hot word and only one device will respond to that hot word.</p><p>Details of one or more embodiments of the subject matter described herein are described by the accompanying drawings and the detailed description of the invention below. Other features, embodiments, and advantages of the subject will become apparent from the detailed description of the invention, the drawings, and the description of the claims.</p>

<figref num="1">It is a figure which shows an example of the system for hot word detection.</figref><figref num="2">It is a figure which shows an example of the process for hot word detection.</figref><figref num="3">It is a figure which shows an example of a computing device and a mobile computing device.</figref>

In the figure, similar symbols refer to similar elements.

In the not too distant future, many devices will always listen to hotwords. If a user has multiple devices trained to respond to his or her voice (eg, phones, tablet computers, TVs, etc.), the user is hot with a device that he or she did not intend to speak to. You would want to suppress the response to the word. For example, if a user issues a hotword to one device and another device of the user is nearby, they will probably also trigger a voice search. In many cases, such behavior is not intended by the user. Therefore, it can be an advantage that only one device, especially the one the user talks to, boots. The present specification deals with the task of selecting the appropriate device to react to the hot word and suppressing the reaction of other devices to the hot word.

FIG. 1 is a diagram showing an example of a system 100 for hotword detection. In general, system 100 illustrates a situation in which user 102 makes an utterance 104 detected by a microphone of computing device 106,108,110. The computing devices 106,108,110 process the utterance 104 to determine the possibility that the utterance 104 contains a hotword. The computing devices 106, 108, 110, respectively, transmit data to each other indicating that the utterance 104 may contain a hot word. The computing devices 106, 108, 110, respectively, compare their data and calculate the highest probability that the utterance 104 contains a hotword, and the computing device initiates speech recognition for the 104 in the utterance. The computing device that did not calculate the highest probability that utterance 104 contains a hotword does not initiate speech recognition for the voice following utterance 104.

Computing devices located close to each other identify each other before the utterance 104 sends data indicating the potential for a hotword to other computing devices. In some embodiments, computing devices identify each other by searching the local network for other devices that are configured to respond to hotwords. For example, the computing device 106 may search the local area network for other devices configured to respond to hotwords to identify the computing device 108 and the computing device 110.

In some embodiments, the computing device identifies another nearby computing device that is configured to respond to a hotword by identifying the user logged in to each device. For example, user 102 is logged in to three computing devices 106,108,110. User 102 holds the computing device 106 in his hand. The computing device 108 is placed on a table and the computing device 110 is located near the wall. The computing device 106 detects the computing devices 108,110, and each computing device shares information about the user logged in to the computing device, such as a user identifier. In some embodiments, the computing device is turned into a hotword by identifying the computing device, which is configured to respond when the hotword is issued by the same user, through speaker identification. It may identify other nearby computing devices that are configured to respond. For example, user 102 responds to his or her own voice when he or she utters a hotword, respectively. Configure 110. Computing devices share speaker identification information by providing each other with user identifiers for user 102. In some embodiments, the computing device may identify other computing devices configured to respond to hotwords via short-range wireless communication. For example, the computing device 106 may transmit signals over short-range wireless communication to search for other computing devices configured to respond to hotwords. Computing devices may utilize one or a combination of these technologies to identify other computing devices that are configured to respond to hotwords.

When computing devices 106,108,110 identify other computing devices that are configured to respond to hotwords, they share and store device identifiers for the identified computing devices. The identifier may be based on the type of device, the IP address of the device, the MAC address, the name given to the device by the user, or any similar uniquely determined identifier. For example, the device identifier 112 for the computing device 106 may be "phone". The device identifier 114 for the computing device 108 may be "tablet". The device identifier 116 for the computing device 110 may be "thermostat". Computing devices 106,108,110 store device identifiers for other computing devices configured to respond to hotwords. Each computing device has a device group that stores the device identifier. For example, the computing device 106 has a device group 118 listing "tablet" and "thermostat" as the two devices it calculates that the audio data will receive the possibility of containing hotwords. The computing device 108 has a device group 120 listing "phone" and "thermostat" as two devices that it calculates and will receive the possibility that the audio data may contain hotwords. The computing device 110 has a device group 122 that lists "phone" and "tablet" as two devices that receive the possibility that the audio data contains hotwords, which it calculates.

When the user 102 makes an utterance 104, i.e. speaks "OK computer", each computing device with a microphone near the user 102 detects and processes the utterance 104. Each computing device detects the utterance 104 via an audio input device such as a microphone. Each microphone provides audio data to an individual audio subsystem. Individual audio subsystems buffer, filter, and digitize audio data. Also, in some embodiments, each computing device may perform termination determinations and speaker identification on the audio data. The audio subsystem provides the processed audio data to the hotword section. The hotword section compares the processed audio data with known hotword data to calculate a reliability score indicating the likelihood that the utterance 104 corresponds to the hotword. The hotword section may extract audio features such as filter bank energy or mel frequency cepstrum coefficients from the processed audio data. The hotword section may use a classification window and process these audio features, such as by using a support vector machine or neural network. Based on the processing of the audio features, the hotword section 124 calculates the reliability score 0.85, the hotword section 126 calculates the reliability score 0.6, and the hotword section 128 calculates the reliability score 0.45. .. In some embodiments, the confidence score may be normalized to the range of zero to one, where higher numbers indicate higher reliability, with utterance 104 containing hot words.

Each computing device sends individual reliability score data packets to other computing devices in the device group. Each reliability score data packet contains an individual reliability score and an individual device identifier for the computing device. For example, the computing device 106 sends a reliability score data packet 130 containing a reliability score of 0.85 and the device identifier "phone" to a computing device in device group 118, i.e., computing devices 108,110. The computing device 108 transmits the reliability score data packet 132 including the reliability score 0.6 and the device identifier "tablet" to the computing device in the device group 120, that is, the computing device 106, 110. The computing device 110 transmits a reliability score data packet 134 containing a reliability score 0.45 and the device identifier "thermostat" to a computing device in the device group 122, that is, computing devices 106, 108.

In some embodiments, the computing device may send a reliability score data packet if the reliability score has reached the hotword score threshold. For example, if the hotword score threshold is 0.5, the computing device 110 will not send the reliability score data packet 134 to other computing devices in device group 122. The computing devices 106,108 will continue to send reliability score data packets 130,132 to the computing devices in device groups 118,120, respectively.

In some embodiments, the computing device that sends the reliability score data packet may send the reliability score data packet directly to another computing device. For example, the computing device 106 may send reliability score data packets to the computing devices 108,110 via short-range wireless communication. The communication protocol used between the two computing devices may be Universal Plug and Play. In some embodiments, the computing device that sends the reliability score data packet may broadcast the reliability score data packet. In this example, the reliability score data packet may be received by a computing device within the device group and by other computing devices. In some embodiments, the computing device that sends the reliability score data packet sends the reliability score data packet to the server, and then the server sends the reliability score data packet to other compute in the device group. It may be sent to the wing device. The server may be located in the local area network of the computing device or in a network accessible via the Internet. For example, the computing device 108 sends a reliability score data packet 132 and a list of computing devices in the device group 120 to the server. The server sends the reliability score data packet 132 to the computing devices 106, 110. In an example where a computing device sends a reliability score data packet to another computing device, the receiving computing device may return a confirmation notification that it has received the reliability score data packet.

Each computing device uses a score comparison unit to compare the hotword reliability scores it receives. For example, the computing device 106 calculates a hotword reliability score of 0.85 and receives hotword reliability scores of 0.6 and 0.45. In this example, the score comparison unit 136 compares the three scores and identifies that the score 0.85 is the highest. For computing devices 108,110, score comparison units 138,140 reach similar conclusions and identify that the score 0.85 corresponding to computing device 106 is the highest.

The computing device that determines that its hotword reliability score is the highest value starts speech recognition for the speech data that follows the utterance of the hotword. For example, the user may say "OK computer" and the computing device 106 may determine that it has the highest hotword reliability score. The computing device 106 will start speech recognition for the audio data received after the hotword. When the user says "call Alice," the computing device 106 will process the utterance and execute the appropriate command. In some embodiments, receiving the hotword may wake the computing device receiving the hotword from sleep. In this example, the computing device with the highest hotword reliability score stays awake, while the other computing device without the highest hotword reliability score is the voice that follows the hotword utterance. Go to sleep without processing data.

As shown in FIG. 1, the score comparison unit 136 identifies that the hotword reliability score corresponding to the computing device 106 is the highest value. Therefore, the device state 142 is "awake". The score comparison unit 138,140 also identifies that the hotword reliability score corresponding to the computing device 106 is the highest. Therefore, the device states 138,140 are "asleep". In some embodiments, the boot state of the computing device may be unaffected. For example, user 102 could have a computing device 106 in his hand while watching a movie on the computing device 108. When the user 102 speaks "OK computer", the computing device 106 initiates speech recognition for the audio data following the hotword by having the highest hotword reliability score. The computing device 108 does not start speech recognition for the audio data following the hotword and continues to play the movie.

In some embodiments, the computing device, which it determines to have the highest hotword reliability score, waits for a certain amount of time before it begins performing speech recognition for the speech following the hotword. In doing so, the computing device that has calculated the highest hotword reliability score is allowed to initiate speech recognition for the voice following the hotword without waiting for a higher hotword reliability score. As an example, the score comparison unit 136 of the computing device 106 receives the hotword reliability scores 0.6 and 0.45 from the computing devices 108 and 110, respectively, and receives the hotword reliability score 0.85 from the hotword unit 124. The computing device 106 has passed 500 milliseconds from the time when the hotword unit 124 calculated the hotword reliability score of the audio data "OK computer" before performing voice recognition for the voice following the hotword. wait. In the example where the score comparison unit receives a higher score, the computing device does not have to wait for a certain amount of time before setting the device state to "sleep". For example, the hotword unit 126 of the computing device 108 calculates a hotword reliability score of 0.6 and receives hotword reliability scores of 0.85 and 0.45. The computing device 108 can set the device state 144 to "sleep" when it receives a hotword reliability score of 0.85. In this case, it is probable that the computing device 108 received the hotword reliability score 0.85 within a specific time after the hotword unit 126 calculated the hotword reliability score 0.6.

In some embodiments, a computing device is specified to allow time for other computing devices to receive its reliability score data packet when it has the highest hotword reliability score. The reliability score data packet may continue to be broadcast for this time. This method would be most appropriate in the case where a computing device does not return a confirmation notification when it receives a reliability score data packet from another computing device. Therefore, if the computing device 106 sends a reliability score data packet 130 to a computing device in device group 118 and receives a confirmation notification before a certain time, such as 500 milliseconds, the hot word The execution of speech recognition for the following speech may be started. In an example where the computing device broadcasts its own reliability score data packet and does not expect confirmation, the computing device receives a hotword reliability score after a certain amount of time, such as 500 milliseconds. You may continue to broadcast your hotword reliability score until you do, whichever comes first. For example, the computing device 110 calculates a hotword reliability score of 0.45 and initiates broadcasting of the reliability score data packet 134. After 300 milliseconds, the computing device 110 receives the reliability score data packet 130 and the hotword reliability score 0.85 from the reliability score data packet 130 is higher than its own hotword reliability score 0.45. Therefore, the broadcast of the reliability score data packet 134 is terminated. In another broadcast example, the compute device 106 has a hotword reliability score of 0. Calculate 85 and start broadcasting the reliability score data packet 130. After 500 milliseconds, the computing device 106 ends the broadcast of the reliability score data packet 130 and begins performing speech recognition for the voice following the hotword. The computing device 106 may receive the reliability score data packet 132,134 before the lapse of 500 milliseconds, but the hotword reliability score in the reliability score data packet 132,134 is lower than 0.85, so that the computing device 106 Continues to wait until 500 milliseconds have passed.

In some embodiments, the computing device may initiate speech recognition for the voice following the hotword before receiving a higher hotword reliability score. The hotword unit calculates the hotword reliability score, and when the hotword reliability score reaches the threshold value, the computing device performs voice recognition for the voice following the hotword. The computing device may perform speech recognition without displaying any indication to the user about speech recognition. This gives users the impression that they can see results based on speech recognition faster than waiting to confirm that they have calculated the highest hotword score, even if the computing device is not active. Given, it would be desirable to do so. As an example, the computing device 106 calculates a hotword reliability score of 0.85 and initiates speech recognition for the voice following the hotword. The computing device 106 receives the reliability score data packets 132,134 and determines that the hotword reliability score 0.85 is the highest value. The computing device 106 continues to perform speech recognition for the speech following the hotword and presents the result to the user. For the computing device 108, the hotword unit 126 calculates a hotword reliability score of 0.6, and the computing device 108 begins performing speech recognition for the voice following the hotword without displaying the data to the user. do. Upon receiving the reliability score data packet 130 including the hotword reliability score 0.85, the computing device 108 ends the execution of speech recognition. No data is displayed to the user, giving the user the impression that the computing device 108 remains in a "sleep" state.

In some embodiments, in order to avoid any waiting time after the hotword is issued, the score is placed prior to the end of the hotword, eg, for a partial hotword, in the hotword section. May be notified from. For example, if the user says "OK computer", the computing device will calculate a partial hotword reliability score when the user finishes saying "OK comp". good. The computing device may then share the partial hotword reliability score with other computing devices. The computing device with the highest partial hotword reliability score can continue to process the user's conversation.

In some embodiments, the computing device emits an audible or inaudible sound, for example, of a particular frequency or frequency pattern, when it determines that the hotword reliability score has reached a threshold. It's okay. The sound will inform other computing devices that the computing device will continue to process the audio data that follows the hotword. Other computing devices will receive this sound and will not process the audio data. For example, the user says "OK computer". One of multiple computing devices calculates a hotword reliability score that is above or equal to the threshold. The computing device emits a sound of 18 kHz when it determines that the hotword reliability score exceeds or is equal to the threshold. Other computing devices near the user may be in the process of calculating the hotword reliability score and may be in the middle of calculating the hotword reliability score when the sound is received. Other computing devices stop processing the user's conversation when they receive the sound. In some embodiments, the computing device may encode the hotword reliability score into audible or inaudible sounds. For example, if the hotword reliability score is 0.5, the computing device may produce audible or inaudible sounds that include a frequency pattern that encodes a score of 0.5.

In some embodiments, the computing device may use different audio measurements to select the computing device to continue processing the user's conversation. For example, a computing device may use loudness to determine which computing device continues to process a user's conversation. The computing device that detects the loudest conversation may continue to process the user's conversation. In another example, a computing device that is currently in use or has an active display may notify other computing devices that it will continue to process the user's conversation when it detects a hotword.

In some embodiments, each computing device near the user when the user is speaking receives audio data and sends the audio data to a server to improve speech recognition. Each computing device can receive audio data corresponding to the user's conversation. Each computing device may send audio data to the server, even though only one computing device appears to the user to handle its conversation. The server can then use the audio data received from each computing device to compare different audio samples corresponding to the same utterance, thus improving speech recognition. For example, the user says, "OK computer, remind me to buy milk (OK computer, remind me to buy milk) ". When the user finishes saying "OK computer", nearby computing devices will probably determine which computing device has the highest hotword reliability score, and that computing device will allow the user to "milk". Remind me to buy a computer, "and it will process those words and return a response. Other computing devices also receive the utterance "Remind me to buy milk." Other computing devices may send audio data to the server that corresponds to "Remind me to buy milk" even if it does not respond to the "Remind me to buy milk" utterance. A computing device that responds to "Remind me to buy milk" may also send its audio data to the server. The server can improve speech recognition because it has different audio samples from different computing devices that process the audio data and respond to the same "remind me to buy milk" utterance. ..

FIG. 2 is a diagram showing an example of the process 200 for hotword detection. Process 200 may be performed by a computing device such as the computing device 108 in FIG. Process 200 calculates a value corresponding to the possibility that the speech contains a hot word, compares that value to another value calculated by another computing device, and puts it in the part of the speech after the hot word. On the other hand, it is determined whether or not to execute voice recognition.

The computing device receives the audio data corresponding to the utterance (210). The user speaks and the microphone of the computing device receives the audio data of the utterance. The computing device processes the audio data by buffering, filtering, endpointing, and digitizing the audio data. As an example, the user speaks "OK computer" and the microphone of the computing device receives the audio data corresponding to "OK computer". The audio subsystem of a computing device samples, buffers, filters, and terminates its audio data for further processing by the computing device.

The computing device determines the first value corresponding to the possibility that the utterance contains a hot word (220). The computing device can be referred to as a hotword reliability score by comparing the audio data of the utterance with a group of audio samples containing the hotword, or by analyzing the audio characteristics of the audio data of the utterance. Determine the value of. The first value may be normalized to the range of zero to one, where 1 indicates the highest likelihood that the utterance will contain a hot word. In some embodiments, the computing device identifies a second computing device so that the second computing device is configured to respond to utterances, including hotwords, and responds to hotwords. Determines that it has been set by the user. The user may log in to both the computing device and the second computing device. Both the computing device and the second computing device may be configured to respond to the user's voice. The computing device and the second computing device may be connected to the same local area network. Both the computing device and the second computing device can be located within a specific distance from each other, such as 10 meters, as determined by GPS or signal strength. For example, the computing device may communicate by short-range wireless communication. The computing device may detect the strength of the signal transmitted by the second computing device, such as 5 dBm, and convert it to a corresponding distance, such as 5 meters.

The computing device receives a second value corresponding to the possibility that the utterance contains a hotword, the second value being determined by the second computing device (230). The second computing device receives the utterance through its own second microphone. The second computing device processes the received audio data corresponding to the utterance to determine a second value or a second hotword reliability score. The second hotword reliability score represents the possibility that the utterance contains a hotword, as calculated by the second computing device. In some embodiments, the computing device uses one or more of the following techniques to transmit a first value to a second computing device. That is, the computing device sets the first value via a server accessible over the Internet, via a server located in a local area network, or directly via a local area network or short-range wireless communication. Can be sent to 2 computing devices. The computing device may send the first value only to the second computing device, or the computing device may receive the first value so that other computing devices can also receive the first value. May be broadcast. The computing device may receive the second value from the second computing device using the same or different technology as the computing device that transmitted the first value.

In some embodiments, the computing device may calculate a volume score for an utterance or a signal-to-noise ratio for an utterance. The computing device may combine the volume score, signal-to-noise ratio, and hotword reliability score to determine new values for comparison with similar values from other computing devices. For example, a computing device may calculate a hotword reliability score and a signal-to-noise ratio. The computing device may then combine these two scores and compare them to scores similarly calculated for other computing devices. In some embodiments, the computing device may calculate different scores and send each score to another computing device for comparison. For example, a computing device may calculate a volume score and a hotword reliability score for an utterance. The computing device may then send these scores to other computing devices for comparison.

In some embodiments, the computing device may transmit a first identifier along with a first value. The identifier may be based on one or more of the addresses of the computing device, the name of the computing device given by the user, or the location of the computing device. For example, the identifier may be "69.123.132.43" or "phone". Similarly, the second computing device may send the second identifier along with the second value. In some embodiments, the computing device may transmit a first identifier to a particular computing device, which is a computing device previously identified as being configured to respond to a hotword. .. For example, a computing device is because the second computing device is capable of responding to hotwords and the same user is logged in to the second computing device in the same way as the computing device. The second computing device may have previously been identified as being configured to respond to hotwords.

The computing device compares the first value with the second value (240). The computing device then initiates speech recognition processing on the audio data based on the results of the comparison (250). In some embodiments, for example, the computing device initiates speech recognition when the first value is greater than or equal to the second value. If the user says "OK computer, call Carol", the computing device will say "OK computer, call Carol" because the first value is greater than or equal to the second value. By performing voice recognition for "Call Carol", the process of "Call Carol" will be started. In some embodiments, the computing device sets the boot state. In an example where the first value is greater than or equal to the second value, the computing device sets the boot state as active or "awake". In the "awake" state, the computing device displays the result of speech recognition.

In some embodiments, the computing device compares the first value with the second value and determines that the first value is less than the second value. The computing device sets the boot state as inactive or "sleep" based on determining that the first value is less than the second value. In the "sleep" state, the computing device does not appear to the user to be activated or to process audio data.

In some embodiments, when the computing device determines that the first value is greater than or equal to the second value, it may wait a certain amount of time before setting the boot state to active. .. A computing device may wait for a specific amount of time to increase the likelihood that it will not receive higher values from other computing devices. The specific time may be fixed or may vary depending on the technology by which the computing device sends and receives values. In some embodiments, the computing device may continue to transmit the first value for a specific time when it determines that the first value is greater than or equal to the second value. By continuing to transmit the first value for a specific time, the computing device increases the probability that the first value will be received by another computing device. In an example where the computing device determines that the first value is less than the second value, the computing device may stop transmitting the first value.

In some embodiments, the computing device may consider additional information in determining whether to execute the command following the hotword. An example of additional information may be the part of the utterance that follows the hot word. Typically, the audio data that follows the hotword is "call Sally", "play Halloween Movie", or "set the temperature to 70 degrees Fahrenheit". Please (set heat to 70) Corresponds to commands for computing devices such as degrees). The computing device can identify a typical device that can handle or can handle that type of request. Typically, a request to call someone will be handled by the telephone device based on typical pre-programmed usage or based on the usage pattern of the user of the device. If the user routinely watches a movie on a tablet, the tablet may handle the request to play the movie. If the thermostat has a temperature control function, the thermostat can handle temperature control.

In order for the computing device to take into account the part of the utterance that follows the hotword, the computing device will have to start speech recognition for the audio data when it would have identified the hotword. The computing device may classify the command portion of the utterance and calculate the frequency of commands in such classification. The computing device may transmit its frequency to other computing devices along with the hotword reliability score. Each computing device may use the frequency and the hotword reliability score to determine whether to execute the command following the hotword.

For example, on a phone where the user says "OK computer, play Michael Jackson" and the computing device is used by the user to listen to music with a 20% chance. If so, the computing device may send that information along with a hotword reliability score. A computing device, such as a tablet, that is used by a user with a 5 percent chance of listening to music may send that information to other computing devices along with a hotword reliability score. The computing device may use the combination of the hotword reliability score and the music reproduction probability to determine whether or not to execute the command.

FIG. 3 shows an example of a computing device 300 and a mobile computing device 350 that can be used to implement the techniques described herein. Computing device 300 is intended to refer to various forms of digital computers such as laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframes, and other suitable computers. .. The mobile computing device 350 is intended to refer to various forms of mobile devices such as personal digital assistants, mobile phones, smartphones, and other similar computing devices. The components shown herein, their connections and relationships, and their functions are for illustrative purposes only and are not meant to be limiting.

The computing device 300 includes a processor 302, a memory 304, a storage device 306, a high-speed interface 308 connecting the memory 304 and a plurality of high-speed expansion ports 310, and a low-speed interface 312 connecting the low-speed expansion port 314 and the storage device 306. include. Each of the processor 302, memory 304, storage device 306, high-speed interface 308, high-speed expansion port 310, and low-speed interface 312 are interconnected using various buses, on a common motherboard or otherwise appropriately. May be mounted. Processor 302 processes instructions for execution within the computing device 300, including instructions stored in memory 304 or on storage device 306, and external inputs such as display 316 connected to high-speed interface 308. / Can display graphical information for the GUI on the output device. In another embodiment, a plurality of processors and / or a plurality of buses may be appropriately used together with a plurality of memories and a plurality of types of memories. Also, multiple computing devices may be connected to each device that provides the required parts of operation (eg, as a server bank, a group of blade servers, or a multiprocessor system).

The memory 304 stores information in the computing device 300. In some embodiments, the memory 304 is one or more volatile memory units. In some embodiments, the memory 304 is one or more non-volatile memory units. Further, the memory 304 may be another form of computer-readable medium such as magnetic or optical disk.

The storage device 306 can provide a large amount of storage for the computing device 300. In some embodiments, the storage device 306 includes a storage area network or other configured device, such as a floppy (registered trademark) disk device, hard disk device, optical disk device, or tape device, flash memory or other similar device. It may be a computer readable medium such as a solid state memory device, or an array of devices, or may include them. The instructions may be stored in the information carrier. When an instruction is executed by one or more processing devices (eg, processor 302), it executes one or more methods, such as those described above. Instructions may also be stored by one or more storage devices, such as a computer-readable or machine-readable medium (eg, memory 304, storage device 306, or memory on processor 302).

The fast interface 308 manages the bandwidth-intensive operation for the computing device 300, while the slow interface 312 manages the lower bandwidth-intensive operation. The allocation of such functions is just one example. In some embodiments, the high-speed interface 308 is in memory 304, in display 316 (eg, via a graphics processor or accelerator), and in high-speed expansion port 310 that can accept various expansion cards (not shown). Be connected. In this embodiment, the slow interface 312 is connected to the storage device 306 and the slow expansion port 314. Slow expansion port 314, which may include various communication ports (eg USB, Bluetooth®, Ethernet®, wireless Ethernet), is a switch or a switch or, for example, via a keyboard, pointing device, scanner, or network adapter. It may be connected to one or more input / output devices, such as networking devices such as routers.

The computing device 300 can be implemented in a number of different forms, as shown in the figure. For example, the computing device 300 may be implemented as including a standard server 320 or a plurality of groups of such servers. Further, the computing device 300 may be implemented as a personal computer such as a laptop computer 322. The computing device 300 may also be implemented as part of the rack server system 324. Alternatively, the components in the computing device 300 may be combined with other components in the mobile device (not shown), such as the mobile computing device 350. Each such device may include one or more of the computing device 300 and the mobile computing device 350, and the entire system may consist of a plurality of computing devices communicating with each other.

The mobile computing device 350 includes, among other components, a processor 352, a memory 364, an input / output device such as a display 354, a communication interface 366, and a transmitter / receiver 368. The mobile computing device 350 may also be provided with a storage device such as a microdrive or other device to provide additional storage. Each of the processor 352, memory 364, display 354, communication interface 366, and transmitter / receiver 368 is interconnected using various buses, some of which are on a common motherboard or appropriately others. It may be mounted by the method.

Processor 352 can execute instructions in mobile computing device 350, including instructions stored in memory 364. Processor 352 may be implemented as a chipset of chips containing separate or multiple analog and digital processors. Processor 352 may be provided for coordination of other components of the mobile computing device 350, such as controlling the user interface, executing applications with the mobile computing device 350, and wireless communication with the mobile computing device 350. ..

The processor 352 may interact with the user via the control interface 358 and the display interface 356 connected to the display 354. The display 354 is, for example, a TFT LCD (Thin-Film-Transistor Liquid Crystal Display) display or an OLED (Organic Light Emitting). Diode) display, or other suitable display technology. The display interface 356 may include suitable circuits for driving the display 354 and presenting graphical and other information to the user. The control interface 358 may receive a command from the user and translate it for submission to the processor 352. In addition, the external interface 362 may provide communication with the processor 352 to allow near-area communication between the mobile computing device 350 and other devices. The external interface 362 may be provided, for example, for wired communication in some embodiments or for wireless communication in some embodiments, and a plurality of interfaces may be used.

The memory 364 stores information in the mobile computing device 350. The memory 364 may be implemented as one or more of a computer-readable medium, one or more volatile memory units, or one or more non-volatile memory units. Further, the extended memory 374 may be provided and connected to the mobile computing device 350 via the extended interface 372. The expansion interface 372 is, for example, SIMM (Single In Line Memory). Module) May include a card interface. The extended memory 374 may provide a separate storage space for the mobile computing device 350, or may store applications or other information for the mobile computing device 350. In particular, the extended memory 374 may include instructions for executing or augmenting the process described above, and may also contain secure information. Thus, for example, the extended memory 374 may be provided as a security module for the mobile computing device 350 and may be programmed with instructions permitting the secure use of the mobile computing device 350. In addition, secure applications may be provided via the SIMM card, along with additional information such as identification information placed on the SIMM card in a non-hackable manner.

The memory may include, for example, flash memory and / or NVRAM memory (Non-Volatile Random Access Memory) as described below. In some embodiments, the instructions are stored in the information carrier. When an instruction is executed by one or more processing devices (eg, processor 352), it executes one or more methods, such as those described above. Instructions are also stored by one or more storage devices, such as one or more computer-readable or machine-readable media (eg, memory 364, extended memory 374, or memory on processor 352). good. In some embodiments, the instruction may be received, for example, as a signal propagated on the transmitter / receiver 368 or the external interface 362.

The mobile computing device 350 may perform wireless communication via the communication interface 366 and may include a digital signal processing circuit, if necessary. The communication interface 366 is, in particular, GSM (Registered Trademark) (Global System for Mobile communications) voice call, SMS (Short Message Service), EMS (Enhanced Messaging Service), or MMS (Multimedia Messaging Service) messaging, CDMA (Code Division Multiple). Access), TDMA (Time Division Multiple Access), PDC (Personal Digital Cellular), WCDMA (Registered Trademark) (Wideband Code Division Multiple Access), CDMA2000, or GPRS (General Packet Radio) It may provide communication under various modes or protocols such as Service). Such communication may be performed, for example, via a transmitter / receiver 368 using a radio frequency. Further, short-range communication may be performed by using Bluetooth (registered trademark), Wi-Fi, or other above-mentioned transmitter / receiver (not shown). In addition, the Global Positioning System (GPS) receiver module 370 may provide the mobile computing device 350 with wireless data for additional navigation and positioning, such data on the mobile computing device 350. Can be used properly by applications running on.

In addition, the mobile computing device 350 may audibly communicate using the audio codec 360, and may receive information emitted by the user and convert it into digital information suitable for use. The audio codec 360 may generate audible sound to the user, such as through a speaker, eg, through the speaker of the handset of the mobile computing device 350. Such sounds may include sounds from voice phone calls, may include recorded sounds (eg, voice messages, music files, etc.), and may be generated by an application running on the mobile computing device 350. It may include the sound to be played.

Mobile Computing Device Device 350 may be implemented in a number of different forms, as illustrated. For example, it may be implemented as a mobile phone device 380. It may also be implemented as a smartphone 382, PDA, or other similar mobile device.

Various embodiments of the systems and techniques described herein include digital electronic circuits, integrated circuits, specially designed application specific integrated circuits (ASICs), computer hardware, firmware, software, and / or theirs. It may be realized as a combination. These various embodiments may include implementation as one or more computer programs executable and / or interpretable on a programmable system including at least one programmable processor. The programmable system may be dedicated or general purpose, to receive data and instructions from the storage system, at least one input device, and at least one output device, and to send data and instructions to them. Is connected to.

These computer programs (also known as programs, software, software applications, or code) include machine language instructions for programmable processors, high-level procedural and / or object-oriented programming languages, and / or assemblies /. It may be implemented by a machine language. As used herein, the terms machine-readable and computer-readable media are machine-language to programmable processors, including machine-readable media that receive machine language instructions as machine-readable signals. Refers to any computer program product, device, and / or device used to provide instructions and / or data (eg, magnetic disk, optical disk, memory, programmable logic device (PLD)). The term machine readable signal refers to any signal used to provide machine language instructions and / or data to a programmable processor.

To provide user interaction, the systems and techniques described herein include display devices for displaying information to the user (eg, a cathode line tube (CRT) or liquid crystal display (LCD) monitor). It may be performed on a computer equipped with a keyboard and a pointing device (eg, a mouse or trackball) capable of providing input to the computer by the user. Other types of devices may be used to provide interaction with the user, for example, the feedback provided to the user may be any form of sensory feedback (eg, visual feedback, auditory feedback, or tactile sensation). It may be feedback), and the input from the user may be received in any form including acoustic, audible, or tactile input.

The systems and technologies described herein include back-end components (eg, data servers), or include middleware components (eg, application servers), or front-end components (eg, user-described in the specification). A computing system that includes (a client computer with a graphical user interface or web browser that can interact with embodiments of the system and technology), or any combination of such backend, middleware, or frontend components. May be carried out as. The components of the system may be interconnected by any form or medium of digital data communication (eg, a communication network). Examples of communication networks include local area networks (LANs), wide area networks (WANs), and the Internet.

The computing system may include a client and a server. The client and server are generally located apart from each other and typically interact over a communication network. The client-server relationship arises from the work of computer programs that run on their own computers and have a client-server relationship with each other.

Although some embodiments have been described in detail above, other modifications can be considered. For example, a client application has been described as accessing a (s) representative stations, but in another embodiment, the (s) representative stations may run on one or more servers. It may be utilized by other applications running on one or more processors. Moreover, the illustrated logic flow does not require the order or order as described to obtain the desired result. In addition, other actions may be provided for the indicated flow, or actions may be removed from the indicated flow. In addition, other components may be added to the indicated system, or components may be removed from the indicated system. Therefore, other embodiments are within the scope of the appended claims.

100 system

102 A user

104 Utterance

106,108,110 Computing devices

112,114,116 Device identifier (ID)

118,120,122 device group

124,126,128 Hot word section

130,132,134 Reliability score data packet

136,138,140 Score comparison section

142,144,146 Device status

300 Computing device

302 Processor

304 memory

306 Storage device

308 High speed interface

310 310 Fast expansion port

312 Slow interface

314 Slow expansion port

316 display

320 server

324 Rack server system

350 Mobile computing device

352 Processor

354 display

356 Display interface

358 Control interface

360 Audio codec

362 External interface

364 memory

366 Communication interface

368 Transmitter / receiver

370 GPS receiver module

372 Extended interface

374 Extended memory

380 Mobile phone

382 smartphone