脑电图记录的对短和弦进行的响应斯坦福数字存储库及readme文件翻译（斯坦福大学2012年数据集）

已剪辑自: https://purl.stanford.edu/js383fs8244
Access conditions
Use and reproduction User agrees that, where applicable, content will not be used to identify or to otherwise infringe the privacy or confidentiality rights of individuals. Content distributed via the Stanford Digital Repository may be subject to additional license and use restrictions applied by the depositor. License This work is licensed under a Creative Commons Attribution 3.0 Unported License

使用和复制用户同意，在适用情况下，内容将不会被用于识别或以其他方式侵犯个人的隐私或保密权利。通过斯坦福数字存储库发布的内容可能受到存储方应用的附加许可和使用限制。本作品是在知识共享署名3.0未移植许可下授权的

Description
Type of resource Software, multimedia Date created 2012

Creators/Contributors
Creator Kaneshiro, Blair Collector Nguyen, Duc T. Primary advisor Berger, Jonathan Principal investigator Suppes, Patrick

Abstract/Contents

This dataset contains scalp-recorded EEG responses from two human participants hearing short chord progressions with expected (tonic) and deviant (dominant, flatted supertonic, or silent) cadential events in keys of C Major, B Major, and F Major (12 chord progressions total).
EEG data are published in Matlab (.mat) format, in two forms.

First, two preprocessed data files (one per participant, each around 300MB), used as input to classification in the Kaneshiro et al. (2012) proceedings paper, each contain 108 epoched trials of response to the fifth chord (cadential event) for each of the 12 stimuli, for a total of 1296 trials of data per participant.

Second, 24 minimally preprocessed EEG recordings (12 per participant, each between 650-811MB) each contain 54 tonic-ending progressions and 9 of each deviant-ending progression for every key, for a total of 648 tonic-ending progressions and 108 of each deviant-ending progression per key per participant.

In addition to EEG data, this dataset includes audio files (.wav format) and a notated score of the stimuli, with mappings to stimulus triggers.

这个数据集包含了头皮记录的两个人类参与者的脑电图响应，他们在C大调、B大调和F大调(共12个和弦进行)的短和弦进行时，听到了预期的(主音)和偏差的(属音、压平的超主音或无声)音调事件。脑电图数据以Matlab (.mat)格式发布，有两种形式。

首先，两个预处理的数据文件(每个参与者一个，每个大约300MB)，在Kaneshiro et al.(2012)的会议论文中用作分类的输入，每个包含了对12个刺激的每个响应第五和弦(韵律事件)的108个划时代的试验。每个参与者总共有1296个试验数据。

其次，24个最低限度预处理的EEC记录(每个参与者12个，每个在650-811MB之间)每个包含54个音调结束的进展，每个键包含9个音调结束的进展，总共648个音调结束的进展，每个参与者每个键包含108个音调结束的进展。

除了脑电图数据外，该数据集还包括音频文件(.wav格式)和刺激的标记分数，以及到刺激筛选器的映射。

Subjects
Subject Electroencephalography (EEG)
Tonal processing
Musical expectation
Center for the Study of Language and Information
Center for Computer Research in Music and Acoustics
Stanford Department of Music Genre Dataset

被试
脑电图(EEG)
色调处理
音乐的期望
语言与信息研究中心
音乐和声学计算机研究中心
斯坦福音乐系

Bibliographic information
Related Publication Blair Kaneshiro, Jonathan Berger, Marcos Perreau Guimaraes, and Patrick Suppes (2012). An Exploration of Tonal Expectation Using Single-Trial EEG Classification. In Proceedings of the 12th International Conference on Music Perception and Cognition and the 8th Triennial Conference of the European Society for the Cognitive Sciences of Music, Thessaloniki, Greece. Related item Kaneshiro et al. (2012) proceedings paper.
Also listed in
View in SearchWorks

EEG-Recorded Responses to Short Chord Progressions
Citation
Blair Kaneshiro, Duc T. Nguyen, Marcos Perreau Guimaraes, Jonathan Berger, and Patrick
Suppes (2015). EEG-Recorded Responses to Short Chord Progressions. Stanford Digital
Repository.
Available at: http://purl.stanford.edu/js383fs8244
Contact: Blair Kaneshiro, blairbo@ccrma.stanford.edu

Blair Kaneshiro, Duc T. Nguyen, Marcos Perreau Guimaraes, Jonathan Berger和PatrickSuppes先生(2015)。
对短和弦进行的脑电图记录的反应。
斯坦福大学的数字存储库。
Available at: http://purl.stanford.edu/js383fs8244
Contact: Blair Kaneshiro, blairbo@ccrma.stanford.edu

Related published work
Kaneshiro B, Berger J, Perreau Guimaraes M, and Suppes P (2012). An Exploration of Tonal
Expectation Using Single-Trial EEG Classification. In Proceedings of the 12th International
Conference on Music Perception and Cognition and the 8th Triennial Conference of the
European Society for the Cognitive Sciences of Music, Thessaloniki, Greece.

金城博，Berger J, Perreau Guimaraes M, and Suppes P(2012)。
调性的探索期望使用单次试验脑电图分类。
在第十二届国际会议论文集中音乐感知与认知大会暨美国音乐协会第8届三年一度会议欧洲音乐认知科学学会，希腊塞萨洛尼基。

Dataset Files

• README.pdf (this document)
Informational document describing the dataset.

• Stimuli_audio.zip
Set of 12 chord progressions used as stimuli for the experiment (.wav format).

• Stimuli_score.pdf
Notated score of all 12 stimuli.

• Stimuli_score_annotated.pdf
Notated score of all 12 stimuli, annotated with trigger numbers of each chord event.

• P1_preprocessed.mat and P2_preprocessed.mat
Preprocessed data classified in the Kaneshiro et al. (2012) proceedings paper.

• P1_a1.mat through P2_d3.mat
Set of 24 minimally preprocessed EEG recordings analyzed in the study.
本研究分析了24组最小预处理脑电图记录。

Data acquisition and experimental paradigm
EEG data were recorded from 128 sensors using the Electrical Geodesics, Inc. (EGI) GES 300
system, sampled at 1 kHz with vertex reference. The experimental paradigm and details of data
acquisition are described in the Kaneshiro et al. (2012) proceedings paper cited above.

脑电图数据记录来自128个传感器使用电气测地线公司(EGI) GES 300系统，采样在1khz与顶点参考。
上面引用的Kaneshiro et al(2012)的论文集中描述了数据采集的实验范式和细节。

Stimulus triggers
Each chord progression was presented as a single, continuous sound file. As a chord
progression played, stimulus triggers were delivered at the onset times of beats 1-5 of the chord
progression. See Stimuli_score_annotated.pdf for chord-to-trigger mappings.
NOTE: We have observed, in other experiments using the stimulus-delivery apparatus used
here, a delay of approximately 67ms between the trigger timestamp and the onset of auditory
stimulus playback. Thus, users of this dataset should be aware that the values in the onsets
variable may need to advanced accordingly to account for such a delay. For the present data,
the timing of subsequent triggers within a given chord progression appear to be correct relative
to the timing of the first trigger, with 625 ms between triggers.

每个和弦进行都以单个连续的声音文件的形式呈现。
当和弦进行时，刺激触发点在和弦进行的第1-5拍开始时释放。
有关从和弦到触发器的映射，请参见标注为pdf的刺激物评分。
注:我们已经观察到。在其他使用本实验中使用的刺激传递装置的实验中，触发时间戳和听觉刺激回放开始之间的延迟约为67毫秒。
因此。此数据集的用户应该知道，onsets变量中的值可能需要相应地提前，以解释这种延迟。
对于目前的数据。在给定的和弦进行中，随后的分流器的计时相对于第一个触发器的计时似乎是正确的，分流器之间的间隔为625毫秒。

Preprocessed EEG files
Preprocessing of files P1_preprocessed.mat and P2_preprocessed.mat is as described in
the Kaneshiro et al. (2012) proceedings paper. Only data corresponding to the fifth beat of each
stimulus (the cadence events) were used for classification analysis. As described in the paper, a
subset of the tonic-cadence responses was selected for classification so that the classification
would involve a balanced number of trials for every stimulus. Trial epoching begins at stimulus
onset (no prestimulus interval).

NOTE: Certain electrodes were excluded from classification analysis for the proceedings paper.
Users are encouraged to identify and remove bad electrodes prior to performing classification.
Participant identifiers of the preprocessed .mat files correspond to identifiers used in the
proceedings paper, and to identifiers used in the minimally preprocessed EEG files in this
dataset.

注:某些电极被排除在会议论文的分类分析之外。
我们鼓励用户在进行分类之前识别并移除坏的电极。
预处理.mat文件的参与者标识符对应于在会议论文中使用的标识符，
对应于在这个数据集中最小预处理的EEG文件中使用的标识符

Each preprocessed .mat file contains the following variables:

• fs: Sampling rate in Hz. For both recordings, fs=62.5.
• sessionID: Identifier of the data recording. sessionID is always the same as the .mat
filename, minus the file extension.
• N: Number of time samples used for each trial. For both recordings, N=39.
• Triggers: 1 x 1296 vector containing the labels of trials used for classification.
• y: Trial-space data matrix containing the trials selected for classification. This is the data
matrix that is ready for input to a classifier. Rows of y correspond to trials, and columns
correspond to concatenated electrodes. Thus, data from trial 1 are found in row 1, from
trial 2 are found in row 2, etc.; data from electrode 1 are found in columns 1:N, from
electrode 2 are found in columns (N+1):2N, etc. This matrix has 1296 rows
(corresponding to elements in the Triggers vector) and 4836 (124*39) columns.

• x: Channel-space data matrix of trials selected for classification. Rows of x correspond
to electrodes, and columns correspond to epoched, concatenated trials. Thus, data from
electrode 1 are found in row 1, from electrode 2 are found in row 2, etc.; data from trial 1
are found in columns 1:N, from trial 2 are in columns (N+1):2N, etc. This matrix has 124
rows and 50544 (1296*39) columns. This matrix corresponds to the elements in the
Triggers vector.

• TS: 2 x 2920 matrix. The first row contains all of the chord 5 event labels, in the order
they were presented during experimental sessions; the second row specifies whether
each event was selected for classification, as in order to present a balanced number of
trials for classification, only one tonic-progression ending per block was used for
analysis. The Triggers vector can be derived from this matrix.

• yAll: Trial-space data matrix of all trials. This data matrix can be used as input to a
classifier, but it does not contain a balanced number of trials for every stimulus. The
rows of yAll correspond to trials, and columns correspond to concatenated electrodes.
Thus, data from trial 1 are found in row 1, from trial 2 are found in row 2, etc.; data from
electrode 1 are found in columns 1:N, from electrode 2 are found in columns (N+1):2N,
etc. This matrix has 2920 rows (corresponding to all elements in the first row of the TS
matrix) and 4836 (124*39) columns.

• xAll: Channel-space data matrix of all trials. Rows of xAll correspond to electrodes, and
columns correspond to epoched, concatenated time samples. Thus, data from trial 1 are
found in columns 1:N, from trial 2 are in columns (N+1):2N, etc. The data matrix has 124
rows and 113880 (2920*39) columns. This matrix corresponds to all elements in the first
row of the TS matrix.

Minimally preprocessed EEG files

Preprocessing of files P1_a1.mat through P2_d3.mat differs from the procedure described in
the 2012 Kaneshiro et al. proceedings paper. For minimally preprocessed data, the following
steps were performed using EGI’s Net Station software:
• Data frames were band-pass filtered between 0.3-50 Hz.
• Filtered data were exported to Matlab cell arrays.

Following this, using Matlab we extracted stimulus triggers and onset times, and anonymized
the data. No artifact rejection or re-referencing has been performed on these data files. All 129
channels of data are retained, including the zero-valued channel 129 representing the vertex
reference electrode.

接下来，我们使用Matlab提取刺激的触发点和开始时间，并将数据匿名化。
没有对这些数据文件执行工件拒绝或重新引用。
所有129通道的数据被保留，包括代表顶点参考电极的零值通道129

Each .mat file is coded as follows:

• {participant identifier}_{session}{recording}.mat
• Participant identifier is P1 or P2 and corresponds to the identifiers used in the
Kaneshiro et al. (2012) proceedings paper.
• Session refers to the day. Recordings were taken over four separate days. Session
dates for P1 spanned 17 days, and for P2 spanned nine days.
• Recording refers to the block in the session. Each session comprised three consecutive
blocks.
• Example: P1_b3.mat is participant 1 (P1), second session (b), third block (3).
Every .mat file contains the following variables:
• fs: Sampling rate in Hz. For all recordings, fs=1000.

• sessionID: Identifier of the data recording. sessionID is always the same as the .mat
filename, minus the file extension.
• xContinuous: Continuous (non-epoched) data frame. xContinuous is a 2D matrix with
129 rows and variable number of columns (channels x total time).

• triggers: Vector of stimulus numbers, in the order they occurred in the recording. Always
a column vector of length 1215.
• onsets: Vector of onset times of triggers, in ms (or time samples, since fs=1000).
Always a column vector of length 1215.

触发器:刺激数字的矢量，按它们在记录中出现的顺序排列。
总是列向量lenath 1215。
onsets:触发器开始时间的向量，单位为ms(或时间样本，因为fs=1000)
总是列向量lenath 1215

本文来自互联网用户投稿，文章观点仅代表作者本人，不代表本站立场，不承担相关法律责任。如若转载，请注明出处。 如若内容造成侵权/违法违规/事实不符，请点击【内容举报】进行投诉反馈！