Back to results
US 12,731,564 B2Granted
Song generation based on a text input
AssigneeMicrosoft Technology Licensing, LLC
InventorsGuang Zhou, Wei Wang, Jiawei Li, Peijun Xia, Xianchao Wu, Lu Yang, Yuanchun Xu
Life of this patent
Estimating the term…
Abstract
The disclosure provides a method and an apparatus for song generation. A text input may be received. A topic and an emotion may be extracted from the text input. A melody may be determined according to the topic and the emotion. Lyrics may be generated according to the melody and the text input. A song may be generated at least according to the melody and the lyrics.
Claims13
A computer-implemented method for song generation, comprising:
- receiving a text input;
- classifying the text input using a trained classifier to extract a topic and an emotion from the text input, wherein the trained classifier applies a multi-dimensional emotion category label corresponding to a vector representation of the text input;
- using a deep neural network model to determine an intention based on the topic, wherein the deep neural network model determines the intention based on pre-trained associations between user text inputs and intentions;
- determining, using processing circuitry, a music style according to the intention and the emotion;
- determining, using processing circuitry, a melody according to the topic, the music style, and the emotion, wherein determining the melody includes:
- determining a chord progression sequence according to the emotion;
- determining a rhyme sequence according to the music style;
- and
- determining the melody according to the rhyme sequence and the chord progression sequence;
- generating lyrics according to the melody and the text input;
- generating a song at least according to the melody and the lyrics;
- and
- outputting the song as a playable audio file or a music sheet comprising the melody and lyrics.
10 more elements
An apparatus for song generation, comprising:
- a receiving module, for receiving a text input;
- a trained classifier to classify the text input, wherein the trained classifier applies a multi-dimensional emotion category label corresponding to a vector representation of the text input;
- an extracting module, for extracting a topic and an emotion from the text input;
- a deep neural network model to determine an intention based on the topic, wherein the deep neural network model determines the intention based on pre-trained associations between user text inputs and intentions;
- a music style determination module, for determining a music style according to the intention and the emotion;
- a melody determination module, for determining a melody according to the topic, the music style, and the emotion, wherein determining the melody includes;
- determining a chord progression sequence according to the emotion;
- determining a rhyme sequence according to the music style;
- and
- determining the melody according to the rhyme sequence and the chord progression sequence;
- a lyrics generation module, for generating lyrics according to the melody and the text input;
- and
- a song generation module, for generating a song at least according to the melody and the lyrics;
- wherein the music style determination module is further to output the song as a playable audio file or a music sheet comprising the melody and lyrics.
11 more elements
An apparatus for song generation, comprising:
- one or more processors;
- and
- a memory storing computer-executable instructions that, when executed, cause the one or more processors to:
- receive a text input;
- classify the text input using a trained classifier to extract a topic and an emotion from the text input, wherein the trained classifier applies a multi-dimensional emotion category label corresponding to a vector representation of the text input;
- determine, using a deep neural network model, an intention based on the topic, wherein the deep neural network model determines the intention based on pre-trained associations between user text inputs and intentions;
- determine a music style according to the intention and the emotion;
- determine a melody according to the topic, the music style, and the emotion, wherein determining the melody includes:
- determining a chord progression sequence according to the emotion;
- determining a rhyme sequence according to the music style;
- and
- determining the melody according to the rhyme sequence and the chord progression sequence;
- generate lyrics according to the melody and the text input;
- generate a song at least according to the melody and the lyrics;
- and
- output the song as a playable audio file or a music sheet comprising the melody and lyrics.
13 more elements
Dependencies are inferred from the claim text.