Text to music selection system
Systems and methods for text-based music selection through a music service may include (1) determining that a user of a social media application may be interested in sharing, via the social media application, music that relates to one or more key words, (2) providing the user with a music recommendation that relates to the one or more key words, (3) receiving input from the user accepting the music recommendation, and (4) in response to receiving the input, sharing the music composition via the social media application. Various other methods, systems, and computer-readable media are also disclosed.
1. A computer-implemented method comprising:
determining that a user of a social media application may be interested in sharing, via the social media application, music that relates to one or more key words;
using a two-tower neural network comprising a text tower and a music tower to identify a plurality of music compositions that relate to the one or more key words by (1) creating a text embedding for the one or more key words, based on one or more features of the key words, and, (2) for each music composition within the plurality, (i) creating a music embedding for the music composition based on one or more musical features of the music embedding, (ii) comparing a closeness of the text embedding with the music embedding, and (iii) identifying the music composition based on a closeness of the music composition's music embedding to the text embedding relative to a closeness of the text embedding to the music embeddings of one or more other music compositions within a group of candidate music compositions;
applying the plurality of music compositions to a ranking system that (1) ranks each music composition within the plurality based on a probability that the user will share the music composition and (2) randomizes the plurality of music compositions;
providing the user with a music recommendation comprising a designated number of the plurality of music compositions that have been selected and ordered based on rankings assigned by the ranking system;
receiving input from the user accepting a music composition from the music recommendation; and
in response to receiving the input, sharing the accepted music composition via the social media application.
2. The computer-implemented method of claim 1 , wherein determining the user's interest in sharing music comprises receiving user input requesting a music sharing recommendation.
3. The computer-implemented method of claim 2 , wherein the user input requesting a music sharing recommendation comprises the one or more key words.
4. The computer-implemented method of claim 1 , wherein determining the user's interest in sharing music comprises:
receiving user input initiating the creation of a social media composition;
deducing that the user may be interested in adding music to the social media composition; and
identifying the one or more key words from text submitted for the social media composition.
5. The computer-implemented method of claim 4 , wherein the social media composition comprises at least one of:
a newsfeed composition;
a digital story composition; or
a private digital message.
6. The computer-implemented method of claim 1 , wherein determining the user's interest in sharing music comprises receiving user input that initiates adding music to a profile of the user within the social media application.
7. The computer-implemented method of claim 6 , wherein the user input that initiates adding music to the profile comprises the one or more key words.
8. The computer-implemented method of claim 1 , wherein the one or more musical features comprise at least one of a tempo, a beat, a music type, or a key.
9. The computer-implemented method of claim 1 , wherein the music embedding for a music composition is further based on one or more lyrical features of the music composition.
10. The computer-implemented method of claim 9 , wherein the lyrical features of the music composition comprise at least one of a word, a set of words, or a phrase included in the music composition.
11. The computer-implemented method of claim 1 , wherein the music embedding for a music composition is further based on an era during which the music composition was created.
12. A system comprising:
a determination module, stored in memory, that determines that a user of a social media application may be interested in sharing, via the social media application, music that relates to one or more key words;
a recommendation module, stored in memory, that:
uses a two-tower neural network comprising a text tower and a music tower to identify a plurality of music compositions that relate to the one or more key words by (1) creating a text embedding for the one or more key words, based on one or more features of the key words, and, (2) for each music composition within the plurality, (i) creating a music embedding for the music composition based on one or more musical features of the music embedding, (ii) comparing a closeness of the text embedding with the music embedding, and (iii) identifying the music composition based on a closeness of the music composition's music embedding to the text embedding relative to a closeness of the text embedding to the music embeddings of one or more other music compositions within a group of candidate music compositions; and
applies the plurality of music compositions to a ranking system that (1) ranks each music composition within the plurality based on a probability that the user will share the music composition and (2) randomizes the plurality of music compositions;
provides the user with a music recommendation comprising a designated number of the plurality of music compositions that have been selected and ordered based on rankings assigned by the ranking system;
an input module, stored in memory, that receives input from the user accepting a music composition from the music recommendation;
a sharing module, stored in memory, that, in response to the input module receiving the input, shares the accepted music composition via the social media application; and
at least one physical processor configured to execute the determination module, the recommendation module, the input module, and the sharing module.
13. The system of claim 12 , wherein the determination module determines the user's interest in sharing music in response to receiving user input requesting a music sharing recommendation.
14. The system of claim 12 , wherein the determination module determines the user's interest in sharing music by:
receiving user input initiating the creation of a social media composition;
deducing that the user may be interested in adding music to the social media composition; and
identifying the one or more key words from text submitted for the social media composition.
15. The system of claim 14 , wherein the social media composition comprises at least one of:
a newsfeed composition;
a digital story composition; or
a private digital message.
16. The system of claim 12 , wherein the determination module determines the user's interest in sharing music in response to receiving user input that initiates adding music to a profile of the user within the social media application.
17. The system of claim 12 , wherein the one or more musical features comprise at least one of a tempo, a beat, a music type, or a key.
18. The system of claim 12 , wherein the music embedding for a music composition is further based on one or more lyrical features of the music composition.
19. The system of claim 18 , wherein the lyrical features of the music composition comprise at least one of a word, a set of words, or a phrase included in the music composition.
20. A non-transitory computer-readable medium comprising one or more computer-readable instructions that, when executed by at least one processor of a computing device, cause the computing device to:
determine that a user of a social media application may be interested in sharing, via the social media application, music that relates to one or more key words;
uses a two-tower neural network comprising a text tower and a music tower to identify a plurality of music compositions that relate to the one or more key words by (1) creating a text embedding for the one or more key words, based on one or more features of the key words, and, (2) for each music composition within the plurality, (i) creating a music embedding for the music composition based on one or more musical features of the music embedding, (ii) comparing a closeness of the text embedding with the music embedding, and (iii) identifying the music composition based on a closeness of the music composition's music embedding to the text embedding relative to a closeness of the text embedding to the music embeddings of one or more other music compositions within a group of candidate music compositions;
apply the plurality of music compositions to a ranking system that (1) ranks each music composition within the plurality based on a probability that the user will share the music composition and (2) randomizes the plurality of music compositions;
provide the user with a music recommendation comprising a designated number of the plurality of music compositions that have been selected and ordered based on rankings assigned by the ranking system;
receive input from the user accepting a music composition from the music recommendation; and
in response to receiving the input, share the accepted music composition via the social media application.