Spotify Ab

Angus William Sackfield

Ching Chuan Sung

Andreas Simon Thore Jansson

Sten Garmark

Dariusz Dziuk

Mateo Rando

Rachel Malia Bittner

Juan José Bosch Vicente

Peter Milan Thomson Sobot

David Rubinstein

Youn Jin Kim

Peter Sobot

A system, method and computer product for training a neural network system. The method comprises inputting an audio signal to the system to generate plural outputs f(X, Θ). The audio signal includes one or more of vocal content and/or musical instrument content, and each output f(X, Θ) corresponds to a respective one of the different content types. The method also comprises comparing individual outputs f(X, Θ) of the neural network system to corresponding target signals. For each compared output f(X, Θ), at least one parameter of the system is adjusted to reduce a result of the comparing performed for the output f(X, Θ), to train the system to estimate the different content types. In one example embodiment, the system comprises a U-Net architecture. After training, the system can estimate various different types of vocal and/or instrument components of an audio signal, depending on which type of component(s) the system is trained to estimate.

Automatic isolation of multiple instruments from musical mixtures

A system, method and computer product for training a neural network system. The method comprises applying an audio signal to the neural network system, the audio signal including a vocal component and a non-vocal component. The method also comprises comparing an output of the neural network system to a target signal, and adjusting at least one parameter of the neural network system to reduce a result of the comparing, for training the neural network system to estimate one of the vocal component and the non-vocal component. In one example embodiment, the system comprises a U-Net architecture. After training, the system can estimate vocal or instrumental components of an audio signal, depending on which type of component the system is trained to estimate.

Singing voice separation with deep U-Net convolutional networks

A system, method and computer product for training a neural network system. The method comprises inputting an audio signal to the system to generate plural outputs f(X, &#x398;). The audio signal includes one or more of vocal content and/or musical instrument content, and each output f(X, &#x398;) corresponds to a respective one of the different content types. The method also comprises comparing individual outputs f(X, &#x398;) of the neural network system to corresponding target signals. For each compared output f(X, &#x398;), at least one parameter of the system is adjusted to reduce a result of the comparing performed for the output f(X, &#x398;), to train the system to estimate the different content types. In one example embodiment, the system comprises a U-Net architecture. After training, the system can estimate various different types of vocal and/or instrument components of an audio signal, depending on which type of component(s) the system is trained to estimate.

A system, method and computer product for combining audio tracks. In one example embodiment herein, the method comprises determining at least one music track that is musically compatible with a base music track, aligning those tracks in time, and combining the tracks. In one example embodiment herein, the tracks may be music tracks of different songs, the base music track can be an instrumental accompaniment track, and the at least one music track can be a vocal track. Also in one example embodiment herein, the determining is based on musical characteristics associated with at least one of the tracks, such as an acoustic feature vector distance between tracks, a likelihood of at least one track including a vocal component, a tempo, or musical key. Also, determining of musical compatibility can include determining at least one of a vertical musical compatibility or a horizontal musical compatibility among tracks.

Method, system, and computer-readable medium for creating song mashups

A system and methods for acquiring cadence and selecting a song version based on the acquired cadence are disclosed. If the system detects a new cadence, then a new song version that corresponds to the new cadence can be played. The new song version playback can start in a corresponding position as the location of playback in a currently-playing song version. Each related song version shares one or more characteristics, such as melody, but is different in at least one characteristic, such as tempo.

Cadence-based selection, playback, and transition between song versions

A system, method and computer product for estimating a component of a provided audio signal. The method comprises converting the provided audio signal to an image, processing the image with a neural network trained to estimate one of vocal content and instrumental content, and storing a spectral mask output from the neural network as a result of the image being processed by the neural network. The neural network is a U-Net. The method also comprises providing the spectral mask to a client media playback device, which applies the spectral mask to a spectrogram of the provided audio signal, to provide a masked spectrogram. The media playback device also transforms the masked spectrogram to an audio signal, and plays back that audio signal via an output user interface.

Growing community of inventors

Stockholm, Sweden

Angus William Sackfield