Systems and Methods for Generating Recommendations in a Digital Audio Workstation

Abstract

A method includes displaying a user interface of a digital audio workstation, which includes a first region for generating a composition. The first region includes a first compositional segment that has been added to the composition by a user. Based on the first compositional segment, one or more recommended predefined compositional segments are identified and displayed in a second region. The method includes receiving the selection of a second compositional segment. The method includes adding the compositional segment to the composition.

1 . (canceled) 
     
     
         2 . A method, comprising:
 at a server system: 
 receiving, from a device that displays a user interface of a digital audio workstation (DAW) for generating a composition, an indication of a first compositional segment that has been added to the composition by a user; 
 identifying, based on the first compositional segment that has already been added to the composition by the user, a first set of one or more recommended predefined compositional segments using a model that is trained on combinations of compositional segments that other users have selected to be included in other compositions; 
 receiving, from the device, an indication of a user selection of a second compositional segment from the first set of one or more recommended predefined compositional segments; and 
 in response to receiving the indication of the user selection of the second compositional segment, adding the second compositional segment to the composition. 
   
     
     
         3 . The method of  claim 2 , further comprising, providing, for display at the device that displays the user interface of the DAW, the first set of one or more recommended predefined compositional segments. 
     
     
         4 . The method of  claim 2 , further comprising, training, at the server system, the model used for identifying the first set of one or more recommended predefined compositional segments, including:
 providing, to a neural network, combinations of compositional segments that other users have included in compositions; and   training the neural network using the combinations of compositional segments to output representations of the compositional segments as vectors in a vector space.   
     
     
         5 . The method of  claim 4 , wherein the neural network is trained using the combinations of compositional segments without regard to content of the compositional segments. 
     
     
         6 . The method of  claim 2 , further comprising:
 in response to receiving the indication of the user selection of the second compositional segment, providing, to the device, an update to a region of the user interface of the DAW to display a second set of one or more recommended predefined compositional segments that are identified based on the first compositional segment and the second compositional segment.   
     
     
         7 . The method of  claim 6 , wherein the second set of one or more recommended predefined compositional segments are identified using the model used for identifying the first set of one or more recommended predefined compositional segments. 
     
     
         8 . The method of  claim 2 , wherein identifying, based on the first compositional segment that has already been added to the composition by the user, the first set of one or more recommended predefined compositional segments includes:
 representing a plurality of compositional segments, including the first set of one or more recommended predefined compositional segments, as respective vectors in a vector space;   generating a first vector using compositional segments, including the first compositional segment, that are present in the composition; and   selecting the first set of one or more recommended predefined compositional segments from the plurality of compositional segments based on vector distances between the first vector and vectors representing respective ones of the plurality of compositional segments.   
     
     
         9 . The method of  claim 8 , wherein generating the first vector using the compositional segments that are present in the composition comprises averaging respective vectors of the compositional segments that are present in the composition. 
     
     
         10 . The method of  claim 8 , further including:
 generating a respective vector corresponding to each respective compositional segment of the plurality of compositional segments by applying, to an input of a neural network, a unique identifier for the respective compositional segment.   
     
     
         11 . The method of  claim 10 , wherein the neural network is a word2vec neural network. 
     
     
         12 . The method of  claim 10 , wherein the unique identifier is not based on content of the respective compositional segment. 
     
     
         13 . The method of  claim 10 , wherein the neural network is trained using data indicating temporally-aligned combinations of compositional segments that other users have included in other compositions. 
     
     
         14 . The method of  claim 10 , wherein the respective vector corresponding to each respective compositional segment of the plurality of compositional segments is characterized by a dimension of at least 50. 
     
     
         15 . A server system, comprising:
 one or more processors;   memory storing one or more programs executable by the one or more processors, the one or more programs including instructions for: 
 receiving, from a device that displays a user interface of a digital audio workstation (DAW) for generating a composition, an indication of a first compositional segment that has been added to the composition by a user; 
 identifying, based on the first compositional segment that has already been added to the composition by the user, a first set of one or more recommended predefined compositional segments using a model that is trained on combinations of compositional segments that other users have selected to be included in other compositions; 
 receiving, from the device, an indication of a user selection of a second compositional segment from the first set of one or more recommended predefined compositional segments; and 
 in response to receiving the indication of the user selection of the second compositional segment, adding the second compositional segment to the composition. 
   
     
     
         16 . The server system of  claim 15 , the one or more programs further comprising instructions for, providing, for display at the device that displays the user interface of the DAW, the first set of one or more recommended predefined compositional segments. 
     
     
         17 . The server system of  claim 15 , the one or more programs further comprising instructions for training, at the server system, the model used for identifying the first set of one or more recommended predefined compositional segments, including:
 providing, to a neural network, combinations of compositional segments that other users have included in compositions; and   training the neural network using the combinations of compositional segments to output representations of the compositional segments as vectors in a vector space.   
     
     
         18 . The server system of  claim 17 , wherein the neural network is trained using the combinations of compositional segments without regard to content of the compositional segments. 
     
     
         19 . The server system of  claim 15 , the one or more programs further comprising instructions for:
 in response to receiving the indication of the user selection of the second compositional segment, providing, to the device, an update to a region of the user interface of the DAW to display a second set of one or more recommended predefined compositional segments that are identified based on the first compositional segment and the second compositional segment.   
     
     
         20 . The server system of  claim 19 , wherein the second set of one or more recommended predefined compositional segments are identified using the model used for identifying the first set of one or more recommended predefined compositional segments. 
     
     
         21 . A non-transitory computer-readable storage medium containing program instructions for causing a server system to perform a method, comprising:
 receiving, from a device that displays a user interface of a digital audio workstation (DAW) for generating a composition, an indication of a first compositional segment that has been added to the composition by a user;   identifying, based on the first compositional segment that has already been added to the composition by the user, a first set of one or more recommended predefined compositional segments using a model that is trained on combinations of compositional segments that other users have selected to be included in other compositions;   receiving, from the device, an indication of a user selection of a second compositional segment from the first set of one or more recommended predefined compositional segments; and   in response to receiving the indication of the user selection of the second compositional segment, adding the second compositional segment to the composition.