Real-Time Transcription and Separation of Drum Recordings Based on NMF Decomposition
نویسندگان
چکیده
This paper proposes a real-time capable method for transcribing and separating occurrences of single drum instruments in polyphonic drum recordings. Both the detection and the decomposition are based on Non-Negative Matrix Factorization and can be implemented with very small systemic delay. We propose a simple modification to the update rules that allows to capture timedynamic spectral characteristics of the involved drum sounds. The method can be applied in music production and music education software. Performance results with respect to drum transcription are presented and discussed. The evaluation data-set consisting of annotated drum recordings is published for use in further studies in the field. Index Terms drum transcription, source separation, nonnegative matrix factorization, spectral processing, audio plug-in, music production, music education
منابع مشابه
Drum Transcription in Polyphonic Music Using Non-Negative Matrix Factorisation
We present a system that is based on the non-negative matrix factorisation (NMF) algorithm and is able to transcribe drum onset events in polyphonic music. The magnitude spectrogram representation of the input music is divided by the NMF algorithm into source spectra and corresponding time-varying gains. Each of these source components is classified as a drum instrument or non-drum sound and a ...
متن کاملItakura-Saito Nonnegative Factorizations of the Power Spectrogram for Music Signal Decomposition
Nonnegative matrix factorization (NMF) is a popular linear regression technique in the fields of machine learning and signal/image processing. Much research about this topic has been driven by applications in audio. NMF has been for example applied with success to automatic music transcription and audio source separation, where the data is usually taken as the magnitude spectrogram of the sound...
متن کاملDrum Transcription Using Partially Fixed Non-Negative Matrix Factorization with Template Adaptation
In this paper, a template adaptive drum transcription algorithm using partially fixed Non-negative Matrix Factorization (NMF) is presented. The proposed method detects percussive events in complex mixtures of music with a minimal training set. The algorithm decomposes the music signal into two dictionaries: a percussive dictionary initialized with pre-defined drum templates and a harmonic dicti...
متن کاملOn Drum Playing Technique Detection in Polyphonic Mixtures
In this paper, the problem of drum playing technique detection in polyphonic mixtures of music is addressed. We focus on the identification of 4 rudimentary techniques: strike, buzz roll, flam, and drag. The specifics and the challenges of this task are being discussed, and different sets of features are compared, including various features extracted from NMF-based activation functions, as well...
متن کاملBeta Process Sparse Nonnegative Matrix Factorization for Music
Nonnegative matrix factorization (NMF) has been widely used for discovering physically meaningful latent components in audio signals to facilitate source separation. Most of the existing NMF algorithms require that the number of latent components is provided a priori, which is not always possible. In this paper, we leverage developments from the Bayesian nonparametrics and compressive sensing l...
متن کامل