Improved musical onset detection with Convolutional Neural Networks

Schlüter, Jan; Böck, Sebastian

doi:10.1109/icassp.2014.6854953

Cited by 171 publications

(164 citation statements)

References 7 publications

Supporting

Mentioning

160

Contrasting

Order By: Relevance

“…fully connected) layer with one output neuron and tanh (hyperbolic tangent) activation function. The neural architecture basically follows the scheme proposed by Schlüter and Böck in [27] with some modifications concerning -apart from the aforementioned regression, replacing classification -mostly the type of nonlinearity of the layers. We agree with [27] that the rectified linear units in the first convolutional layer may play the role of the half-wave rectifier H function (cf.…”

Section: A Neural Network Architecturementioning

confidence: 99%

“…For the tanh activation function the optimal threshold value T opt determined in our tests, i.e. the value maximizing the F-measure [27][10] was always lower than zero. After the thresholding, the peak-picking procedure is applied and peaks found within the range of 50ms relative to the actual onsets are treated as the properly detected ones.…”

Section: Fig 1 Spectrogram Fragment Enlarged (In a Black Box Lowermentioning

confidence: 99%

“…In this work we adopted the approach proposed in [27] to test the effectiveness of a convolutional network in the onset detection task using two different signal representations, namely the logarithmically scaled spectrogram and enhanced autocorrelation (EAC).…”

Section: Previous Workmentioning

confidence: 99%

“…Apart from bidirectional long shortterm memory neural networks (LSTM) [23] and recurrent neural networks (RNN) [21], the convolutional neural networks (CNNs) [26][27] proved to be especially useful here.…”

Section: Previous Workmentioning

confidence: 99%

See 3 more Smart Citations

Analysis of time-frequency representations for musical onset detection with convolutional neural network.

Stasiak¹,

Mońko²

2016

Annals of Computer Science and Information Systems

View full text Add to dashboard Cite

Abstract-In this paper a convolutional neural network is applied to the problem of note onset detection in audio recordings. Two time-frequency representations are analysed, showing the superiority of standard spectrogram over enhanced autocorrelation (EAC) used as the input to the convolutional network. Experimental evaluation is based on a dataset containing 10,939 annotated onsets, with total duration of the audio recordings of over 45 min.

show abstract

Section: A Neural Network Architecturementioning

confidence: 99%

Section: Fig 1 Spectrogram Fragment Enlarged (In a Black Box Lowermentioning

confidence: 99%

Section: Previous Workmentioning

confidence: 99%

Section: Previous Workmentioning

confidence: 99%

See 2 more Smart Citations

Analysis of time-frequency representations for musical onset detection with convolutional neural network.

Stasiak¹,

Mońko²

2016

Annals of Computer Science and Information Systems

View full text Add to dashboard Cite

show abstract

“…For instance, a multi-net approach proposed by Lacoste and Eck (2007) is based on merging the results obtained from several networks, each trained with a different set of hyper-parameters, by means of an additional "output" neural network followed by a peak-picking procedure. Apart from the standard questions regarding the number of hidden layers and hidden neurons, several different NN types, including the recurrent neural network (RNN), the feed-forward convolutional neural network (CNN) and the LSTM (long short-term memory) neural network, have been considered (Böck et al, 2012;Eyben et al, 2010;Schlüter and Böck, 2014).…”

Section: 3mentioning

confidence: 99%

Note onset detection in musical signals via neural–network–based multi–ODF fusion

Stasiak

Mońko

Niewiadomski

2016

International Journal of Applied Mathematics and Computer Science

View full text Add to dashboard Cite

The problem of note onset detection in musical signals is considered. The proposed solution is based on known approaches in which an onset detection function is defined on the basis of spectral characteristics of audio data. In our approach, several onset detection functions are used simultaneously to form an input vector for a multi-layer non-linear perceptron, which learns to detect onsets in the training data. This is in contrast to standard methods based on thresholding the onset detection functions with a moving average or a moving median. Our approach is also different from most of the current machinelearning-based solutions in that we explicitly use the onset detection functions as an intermediate representation, which may therefore be easily replaced with a different one, e.g., to match the characteristics of a particular audio data source. The results obtained for a database containing annotated onsets for 17 different instruments and ensembles are compared with state-of-the-art solutions.

show abstract

Temporal Analysis

2022

An Introduction to Audio Content Analysis

View full text Add to dashboard Cite

Improved musical onset detection with Convolutional Neural Networks

Cited by 171 publications

References 7 publications

Analysis of time-frequency representations for musical onset detection with convolutional neural network.

Analysis of time-frequency representations for musical onset detection with convolutional neural network.

Note onset detection in musical signals via neural–network–based multi–ODF fusion

Temporal Analysis

Contact Info

Product

Resources

About