Content playback system
The invention provides a content playback system comprising a playback device that is configured to detect a voice command from a user and to play content. When a voice command is received, the system is configured to analyse the voice command to determine a user intent. The system then extracts one or more entities from the voice command, wherein each of the extracted entities is of a type associated with the determined user intent. Then, based on the one or more extracted entities, the system controls the playback device. Analysis of the voice command in this manner may improve an accuracy with which a meaning of the voice command can be obtained, thereby facilitating control of the playback device.
1. A content playback system comprising:
a playback device, the playback device being configured to detect a voice command from a user and to play content;
wherein the system is configured to:
analyse a voice command detected by the playback device to determine a user intent;
analyse the voice command to extract one or more entities from the voice command, wherein each of the extracted entities is of a type associated with the determined user intent;
select an entity from the one or more extracted entities based on a set of conflict resolution rules, wherein the set of conflict resolution rules are set such that, when the one or more extracted entities includes two or more overlapping entities, the longest entity of the overlapping entities is selected; and
control the playback device based on the selected entity.
2. A content playback system according to claim 1 , wherein the system is further configured to store a plurality of predefined user intents, and wherein the user intent is determined from the plurality of predefined user intents.
3. A content playback system according to claim 2 , wherein the system is further configured to:
store, for each of the plurality of predefined user intents, a pattern associated with that predefined user intent; and
determine that a predefined user intent is the user intent if a phrase in the voice command matches the pattern associated with that predefined user intent.
4. A content playback system according to claim 2 , wherein the plurality of predefined user intents is stored as an ordered list, and wherein the system is configured to iteratively determine whether a predefined user intent is the user intent in the order in which the plurality of predefined user intents is stored.
5. A content playback system according to claim 2 , wherein the system is further configured to store, for each of the plurality of predefined user intents, one or more entity types associated with that predefined user intent.
6. A content playback system according to claim 5 , wherein the system is further configured to:
for a first entity type stored by the system, store a plurality of regular expressions associated with the first entity type; and
extract an entity of the first entity type from the voice command by matching a phrase in the voice command with one of the plurality of regular expression associated with the first entity type.
7. A content playback system according to claim 5 , wherein the system is further configured to:
for a second entity type stored by the system, store a phrase structure associated with the second entity type; and
extract an entity of the second type from the voice command by matching a phrase in the voice command with the phrase structure associated with the second entity type.
8. A content playback system according to claim 7 , wherein the phrase structure has a song field and an artist field.
9. A content playback system according to claim 8 , wherein the system is further configured to:
determine an artist name from the phrase in the voice command, the artist name being a string in the phrase that is at a position corresponding to the artist field in the phrase structure;
obtain from a database a list of song names associated with the artist name; and
extract the entity of the second type by matching a song name in the list of song names with a string in the phrase that is at a position corresponding to the song field.
10. A content playback system according to claim 1 , wherein the set of conflict resolution rules are set such that, when the one or more extracted entities includes two or more overlapping entities, one of the overlapping entities having a prioritized entity type is selected.
11. A content playback system according to claim 1 , wherein selecting an entity includes identifying word boundaries in the voice command, and discarding entities which do not start and end at identified word boundaries.
12. A content playback system according to claim 1 , wherein controlling the playback device includes playing content via the playback device.
13. A content playback system according to claim 1 , wherein controlling the playback device includes generating or adapting a content playlist, and playing the content playlist via the playback device.
14. A content playback system according to claim 1 , further comprising a content server configured to store content that is playable by the playback device.
15. A content playback system according to claim 1 , further comprising a controller separate from the playback device, the controller being configured to control the playback device.
16. A content playback system according to claim 1 , wherein the playback device comprises a speaker for playing audio content.
17. A method of controlling a playback device, the playback device being configured to detect a voice command from a user and to play content, the method comprising:
analysing a voice command detected by the playback device to determine a user intent;
analysing the voice command to extract one or more entities from the voice command, wherein each of the extracted entities is of a type associated with the determined user intent;
selecting an entity from the one or more extracted entities based on a set of conflict resolution rules, wherein the set of conflict resolution rules are set such that, when the one or more extracted entities includes two or more overlapping entities, the longest entity of the overlapping entities is selected; and
controlling the playback device based on the selected entity.
18. A method according to claim 17 , further comprising storing a plurality of predefined user intents, wherein the user intent is determined from the plurality of predefined user intents.
19. A method according to claim 18 , further comprising:
storing, for each of the plurality of predefined user intents, a pattern associated with that predefined user intent; and
determining that a predefined user intent is the user intent if a phrase in the voice command matches the pattern associated with that predefined user intent.