MID-LEVEL FEATURES FOR AUDIO CHORD RECOGNITION USING A DEEP NEURAL NETWORK

Research output: Contribution to journalArticleResearchpeer-review

Abstract

Deep neural networks composed of several pre-trained layers have been successfully applied to various tasks related to audio processing. Some configurations of deep neural networks (including deep recurrent networks) which can be pretrained with the help of stacked denoising autoencoders are proposed and examined in this paper in application to feature extraction for audio chord recognition task. The features obtained from an audio spectrogram using such network can be used instead of conventional chroma features to recognize the actual chords in the audio recording. Chord recognition quality that was achieved using the proposed features is compared to the one that was achieved using conventional chroma features which do not rely on any machine learning technique
Original languageEnglish
Pages (from-to)109-117
Number of pages9
JournalУЧЕНЫЕ ЗАПИСКИ КАЗАНСКОГО УНИВЕРСИТЕТА. СЕРИЯ: ФИЗИКО-МАТЕМАТИЧЕСКИЕ НАУКИ
Volume155
Issue number4
Publication statusPublished - 2013

Fingerprint

Audio recordings
Learning systems
Feature extraction
Processing
Deep neural networks

GRNTI

  • 50.00.00 AUTOMATION. COMPUTER ENGINEERING

Level of Research Output

  • VAK List

Cite this

@article{c8058c67186c466286e32d0d475549e7,
title = "MID-LEVEL FEATURES FOR AUDIO CHORD RECOGNITION USING A DEEP NEURAL NETWORK",
abstract = "Deep neural networks composed of several pre-trained layers have been successfully applied to various tasks related to audio processing. Some configurations of deep neural networks (including deep recurrent networks) which can be pretrained with the help of stacked denoising autoencoders are proposed and examined in this paper in application to feature extraction for audio chord recognition task. The features obtained from an audio spectrogram using such network can be used instead of conventional chroma features to recognize the actual chords in the audio recording. Chord recognition quality that was achieved using the proposed features is compared to the one that was achieved using conventional chroma features which do not rely on any machine learning technique",
author = "Nikolai Glazyrin",
year = "2013",
language = "English",
volume = "155",
pages = "109--117",
journal = "УЧЕНЫЕ ЗАПИСКИ КАЗАНСКОГО УНИВЕРСИТЕТА. СЕРИЯ: ФИЗИКО-МАТЕМАТИЧЕСКИЕ НАУКИ",
issn = "2541-7746",
publisher = "Казанский (Приволжский) федеральный университет",
number = "4",

}

TY - JOUR

T1 - MID-LEVEL FEATURES FOR AUDIO CHORD RECOGNITION USING A DEEP NEURAL NETWORK

AU - Glazyrin, Nikolai

PY - 2013

Y1 - 2013

N2 - Deep neural networks composed of several pre-trained layers have been successfully applied to various tasks related to audio processing. Some configurations of deep neural networks (including deep recurrent networks) which can be pretrained with the help of stacked denoising autoencoders are proposed and examined in this paper in application to feature extraction for audio chord recognition task. The features obtained from an audio spectrogram using such network can be used instead of conventional chroma features to recognize the actual chords in the audio recording. Chord recognition quality that was achieved using the proposed features is compared to the one that was achieved using conventional chroma features which do not rely on any machine learning technique

AB - Deep neural networks composed of several pre-trained layers have been successfully applied to various tasks related to audio processing. Some configurations of deep neural networks (including deep recurrent networks) which can be pretrained with the help of stacked denoising autoencoders are proposed and examined in this paper in application to feature extraction for audio chord recognition task. The features obtained from an audio spectrogram using such network can be used instead of conventional chroma features to recognize the actual chords in the audio recording. Chord recognition quality that was achieved using the proposed features is compared to the one that was achieved using conventional chroma features which do not rely on any machine learning technique

UR - http://elibrary.ru/item.asp?id=22002751

M3 - Article

VL - 155

SP - 109

EP - 117

JO - УЧЕНЫЕ ЗАПИСКИ КАЗАНСКОГО УНИВЕРСИТЕТА. СЕРИЯ: ФИЗИКО-МАТЕМАТИЧЕСКИЕ НАУКИ

JF - УЧЕНЫЕ ЗАПИСКИ КАЗАНСКОГО УНИВЕРСИТЕТА. СЕРИЯ: ФИЗИКО-МАТЕМАТИЧЕСКИЕ НАУКИ

SN - 2541-7746

IS - 4

ER -