Invented by Michael J. Lebeau, John Nicholas Jitkoff, William J. Byrne, Google LLC
The Google LLC invention works as followsIn general the subject matter described here can be embodied as methods, systems and program products that receive a voice query on a mobile computing system and generate data to represent the content of the voice request. The data is sent to a server. The mobile computing device receives a textual query determined by the speech recognizer to be the textual form of a part of the data. The textual question is found to contain a carrier phrase consisting of one or several words reserved by the first third-party program installed on a computing device. First, a group of third-party applications is selected to receive part or all of the textual question. The selected first application program receives all or part of the textual question.
Background for Routing Queries based on carrier phrases registration
The user of a mobile computer device, such as a smartphone or application telephone, can download applications to the device via a network-accessible application distribution platform, and then run them on the device. Applications can come in a wide variety of formats and be offered by third parties via an online application marketplace.
The mobile device can also allow users to input information orally. A user could, for example, press a button while speaking the query “San Francisco Pizza Places?” You can view the search results for pizza places in San Francisco by speaking a query. This oral input can speed up a user’s input of data to the device, compared to typing it in. It can also allow the user to use the device even if they are unable to type.
This document describes methods, systems and mechanisms for routing requests based on carrier phrases registration on mobile computing devices. Third-party applications (i.e. programs that are provided by individuals or companies that are not the same as the operating system and provider of the mobile device) can register carrier phrases on a device. Carrier phrases can be single words or words in groups that are translated to text and then used as triggers for commands on the device. When a mobile computing device user orally speaks a query that contains a carrier phrase registered by a third party application program, the query is sent to the application program or actions are taken based on the content of the question.
In general, an aspect of the subject described in this specification may be embodied as a computer program that is tangibly embedded in a computer readable storage medium, and that contains instructions that, when executed by a CPU, perform operations. The operations include receiving a voice request orally provided by a computing system, and generating data to represent the content of that voice query. The mobile computing device can provide the data to the server system. The operation includes receiving, both by the computing system and the server system, the textual query which has been determined to be the textual form of the data by a server system speech recognizer. The operations also include determining whether the textual question includes a carrier term of one or multiple words reserved by a third-party program installed on a computing device. The operations include choosing the first third party application from a list of several third party applications to receive all or part of the textual question. The computing device provides all or part of the textual question to the first application program selected.
Another aspect described in this specification may be implemented in a computer-implemented technique. The method involves receiving, by a computer system, a written query which is a textual version of a spoken query. A speech recognition system determined the textual query from audio data representing the voice query. The server system determines, using the textual question, whether the carrier phrase of words is one or more that has been registered by a third-party program. The plurality of third party application programs have each registered one or several carrier phrases as a response to requests by or on behalf the respective third-party program. The method involves providing a remainder of the textual question to the first application program on the computing device. The remaining textual search is the portion of textual search that does not contain the carrier phrase.
In a third aspect, subject matter described herein can be implemented in a computer-implemented systems. The system includes an computer-implemented voice recognizer configured to (i), receive data representing a content provided orally to a computing unit, and (ii), generate a query textual that is a representation of the data. The system includes a database of carrier phrases which is programmed to identify third-party applications provided by different organizations from the organization that provides this system. It will also identify carrier phrases assigned to each of these third-party applications in response to requests made by the third party application programs and their providers to reserve respective carrier phrases. The application programs can be installed onto the computing device. The system includes an computer-implemented “query distributor” programmed to identify a carrier phrase in the textual request from a plurality.
These and other implementaions can optionally include some or all of the following features. The carrier phrase may be excluded from the entire or part of the textual question that is sent to the first application program. The data can be an audio recording from the voice query that was provided orally. The computing device may have received the voice query after the user selected a button provided by the device to submit voice queries. The textual query can include words in a language that is understandable by humans. The computing device may provide the carrier phrase to the first application program of the third party.
The first third party application program may receive all or part of the query text to perform certain actions. These actions will be defined by the third party application program. They will also be based on the content of all or part of this textual query. Operation may include determining whether third-party applications other than the first program have not registered the carrier phrase included in the query. The textual query may not have been sent to all or part of each of the third-party applications programs. The third-party applications may have been downloaded to the mobile computing device via an install file. The mobile computing device may have received the install files for each third-party program over the internet in response to user input requesting receipt of the file over the Internet.
Determining whether the textual question includes the carrier phrase registered by the first program of application may include comparing the beginning portion of the query with the carrier phrases that have been registered by each of a plurality of programs of application. The plurality may consist of third-party applications that are installed on the mobile computing devices. The plurality third-party applications programs may include web applications programs that the mobile device has visited.
The entire textual question may be any portion of the query that does include the carrier phrase. The system can include a computer implemented voice query encoder that generates data based on a voice query provided orally to the mobile computing devices by the user. The query distributor can be part of the operating system on a mobile computing device. The query distributor cannot provide any third-party applications that are not first-party applications for all or part of the textual queries.
The first application program of a third party may decide what operations to perform on the basis of a textual query that is all or part of. Operating system may not determine the actions to be performed by the first third-party application program. Instructions included in an install file of the first third party application program could define the operations that the first application program should perform. The first application’s install file may have been sent to the computing device via a network as a result of user input requesting the installation file.
The speech recognition system may generate the carrier phrase using a first model of language.” The speech recognizer can select a second language from a number of language models assigned to carrier phrase based on the carrier phrase. Using the second language models, the speech recognizer can generate, based upon the voice query, the remainder of the first textual question that does not contain the carrier phrase. The query distributor can identify carrier phrases by comparing the beginning portions of textual questions to carrier phrases. The query distributor can indicate that the carrier phrases have not been identified if the query distributer does not find carrier phrases at the beginning of textual questions. Third-party applications programs can be web applications loaded by the mobile computing devices upon visiting Uniform Resource Indicators.
In certain cases, “Specific embodiments” can be used to achieve one or more of the advantages listed below. The oral communication interface of a mobile device may expand as third-party apps are installed. Third-party applications can designate their own “carrier phrases” and thus the communication interface. Both application developers and users may benefit from the increased flexibility of the interface. This flexibility could encourage developers to use a computing platform in a more creative way, and users to have a better experience with their device.
The drawings and description below provide details about one or more embodiments. The description, drawings and claims will reveal other features, advantages and objects.
DESCRIPTION of Drawings
FIG. 4 illustrates remote carrier phrase registration.
Click here to view the patent on Google Patents.