A music beat point detection method includes: performing a frame processing on a music signal to obtain a frame signal; obtaining a power spectrum of the frame signal; performing sub-band decomposition on the power spectrum, and decomposing the power spectrum into at least two sub-bands; performing a time-frequency domain joint filtering on a signal of each sub-band according to a beat type corresponding to each sub-band; obtaining a to-be-confirmed beat point from the frame signal of the music signal according to a result of the time-frequency domain joint filtering; and obtaining a beat point of the music signal according to a power value of the to-be-confirmed beat point.

According to an embodiment, there is provided a system and method for automatically generating a complete music work from a partially completed work provided by a user. One approach uses an artificial intelligence (AI) engine that is trained by creating incomplete works from a database of complete works and then instructing the AI to complete the incomplete works. A comparison is made between the completed works and the originals to determine the effectiveness of the training process. After the AI is trained, it is applied to the user's incomplete work to produce a final music item.

According to an embodiment, there is provided a system and method for automatic AI controlled loop based song construction. It provides and benefits from a machine learning AI in a audio loop selection engine for the generation of a song structure and for the selection of fitting audio loops from a database of audio loops. In one embodiment, the instant method provides a music generation process that utilizes an AI system that has been trained and validated on a music item database to complete the creation of a music item given an incomplete song that was started but not finished by a user.

A vehicle engine sound control system identifies a vehicle driver by a driver smartphone or a driver biometric information detecting sensor and analyzes the music to which the identified driver listens with the driver smartphone or a vehicle infotainment system. A traveling pattern of the driver is analyze by applying any one among a vehicle, a GPS, a road, and weather as a condition. A driver propensity engine sound pattern is generated as a result value by learning at least any one information among a driver identifying unit, a music analyzing unit, and a travel analyzing unit. The engine sound is adjusted and output based the result value.

Apparatus and methods for audio classifying and processing are disclosed. In one embodiment, an audio processing apparatus includes an audio classifier for classifying an audio signal into at least one audio type in real time; an audio improving device for improving experience of audience; and an adjusting unit for adjusting at least one parameter of the audio improving device in a continuous manner based on the confidence value of the at least one audio type.

Digital signal processing and machine learning techniques can be employed in a vocal capture and performance social network to computationally generate vocal pitch tracks from a collection of vocal performances captured against a common temporal baseline such as a backing track or an original performance by a popularizing artist. In this way, crowd-sourced pitch tracks may be generated and distributed for use in subsequent karaoke-style vocal audio captures or other applications. Large numbers of performances of a song can be used to generate a pitch track. Computationally determined pitch trackings from individual audio signal encodings of the crowd-sourced vocal performance set are aggregated and processed as an observation sequence of a trained Hidden Markov Model (HMM) or other statistical model to produce an output pitch track.

Methods, systems, and apparatus, including computer programs encoded on a computer storage medium, for speech recognition using complex linear projection are disclosed. In one aspect, a method includes the actions of receiving audio data corresponding to an utterance. The method further includes generating frequency domain data using the audio data. The method further includes processing the frequency domain data using complex linear projection. The method further includes providing the processed frequency domain data to a neural network trained as an acoustic model. The method further includes generating a transcription for the utterance that is determined based at least on output that the neural network provides in response to receiving the processed frequency domain data.

Implementations generally relate to automated equalizer adjustments based on audio characteristics. In some implementations, a method includes detecting music that is being currently played on an audio device. The method further includes adjusting one or more equalizer settings of an equalizer device based at least in part on a music genre associated with the music. The method further includes outputting the music based at least in part on the adjusting of the one or more equalizer settings of the equalizer device.

Systems and methods are provided for enhancements for musical composition applications. An example method includes receiving information identifying initiation of a music composition application, the music composition application being executed via a user device of a user, with the received information indicating a genre associated with a musical score being created via the music composition application. One or more constraints associated with the genre are determined, with the constraints indicating one or more features learned based on analyzing music associated with the genre. Musical elements specified by the user are received via the music composition application. Musical score updates are determined based on the musical elements and genre. The determined musical score updates are provided to the user device.

Methods, systems, and apparatus, including computer programs encoded on a computer storage medium, for speech recognition using complex evolution recurrent neural networks. In some implementations, audio data indicating acoustic characteristics of an utterance is received. A first vector sequence comprising audio features determined from the audio data is generated. A second vector sequence is generated, as output of a first recurrent neural network in response to receiving the first vector sequence as input, where the first recurrent neural network has a transition matrix that implements a cascade of linear operators comprising (i) first linear operators that are complex-valued and unitary, and (ii) one or more second linear operators that are non-unitary. An output vector sequence of a second recurrent neural network is generated. A transcription for the utterance is generated based on the output vector sequence generated by the second recurrent neural network. The transcription for the utterance is provided.