Unsupervised Analysis and Generation of Audio Percussion Sequences

Marchini, Marco; Purwins, H.‐G.

doi:10.1007/978-3-642-23126-1_14

Cited by 5 publications

(9 citation statements)

References 7 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…While Information Gain was shown to be a good indicator of agreement between beat sequences from among existing beat tracking evaluation methods, it is not the only approach which could be used. In this paper we also explore an alternative mechanism for measuring agreement, the regularity function of Marchini and Purwins [35], which quantifies the degree of temporal regularity between time events. To calculate the regularity we first concatenate and sort the beats of two different beat sequences, then we compute the histogram of the time differences between all possible combinations of two beats (the complete inter-beat interval histogram, CIBIH).…”

Section: Selection Methods and Measuring Mutual Agreementmentioning

confidence: 99%

See 1 more Smart Citation

Multi-Feature Beat Tracking

Zapata

Davies

2014

IEEE/ACM Trans. Audio Speech Lang. Process.

View full text Add to dashboard Cite

A recent trend in the field of beat tracking for musical audio signals has been to explore techniques for measuring the level of agreement and disagreement between a committee of beat tracking algorithms. By using beat tracking evaluation methods to compare all pairwise combinations of beat tracker outputs, it has been shown that selecting the beat tracker which most agrees with the remainder of the committee, on a song-by-song basis, leads to improved performance which surpasses the accuracy of any individual beat tracker used on its own. In this paper we extend this idea towards presenting a single, standalone beat tracking solution which can exploit the benefit of mutual agreement without the need to run multiple separate beat tracking algorithms. In contrast to existing work, we re-cast the problem as one of selecting between the beat outputs resulting from a single beat tracking model with multiple, diverse input features. Through extended evaluation on a large annotated database, we show that our multi-feature beat tracker can outperform the state of the art, and thereby demonstrate that there is sufficient diversity in input features for beat tracking, without the need for multiple tracking models.

show abstract

Section: Selection Methods and Measuring Mutual Agreementmentioning

confidence: 99%

“…Given the sequence of beats we define the regularity of the sequence of beats to be: (12) If the beat estimations are more equally spaced in time the regularity value will be higher, whereas if the beat estimations are unrelated the regularity value will be lower. For more information see [35].…”

Section: Selection Methods and Measuring Mutual Agreementmentioning

confidence: 99%

Multi-Feature Beat Tracking

Zapata

Davies

2014

IEEE/ACM Trans. Audio Speech Lang. Process.

View full text Add to dashboard Cite

show abstract

“…Marchini [Mar10] demonstrated a similar system in the context of generating rhythmic variations of a percussion track audio. This system was flexible and robust to audio recorded from several percussive sources such as drums, beat-box, etc.…”

Section: Segmentationmentioning

confidence: 99%

“…As a continuation of what has been introduced in Chapter 1, this chapter provides a theoretical background of stochastic processes, Markov chains and their application to this work. The following description closely follows that of [Mar10].…”

Section: Generationmentioning

confidence: 99%

“…On the other hand, a first order Markov chain is often susceptible to produce seemingly random predictions. As a result, many of the recent approaches have adopted the variable-length Markov Chains (VLMCs), which provide a trade-off between the two extremes, for representing musical information [Mar10], [CW95], [Pac03]. At the same time, they can be represented with a parsimonious, efficient tree structure.…”

mentioning

confidence: 99%

See 1 more Smart Citation

Automatic Phrase Continuation from Guitar and Bass Guitar Melodies

Cherla

Purwins

Marchini

2013

Computer Music Journal

Self Cite

View full text Add to dashboard Cite

A framework is proposed for generating musically similar and interesting variations of a given monophonic melody. In this work, the focus is on rock/pop guitar and bass-guitar melodies with the aim of extensions to other instruments and musical styles. The original audio melody in audio format is first segmented into its component notes using onset detection and pitch estimation. Clustering is performed on the pitches to obtain a symbolic representation of pitch sequences in it. The note onsets are aligned with the estimated meter of the melody for a time-homogeneous symbolization of the rhythm in terms of onsets/rests and the metrical locations of their occurrence. A joint representation based on the cross-product of these two individual representations is used to train the prediction framework -the variablelength Markov chain. It is hypothesized that such a model will rearrange, while maintaining certain coherence, the segments of the original melody. The musical quality of the generated melodies was evaluated through a questionnaire by a group of experts and received an overall positive response.

show abstract