dcase_util教程(二)——各单元介绍

接着上一篇教程,继续的有各个UTILITIES的介绍。网址

1. Container

数据容器的类。这些数据的目的是为了包装数据,使用有用的方法来访问和操作数据,以及加载和存储数据。

这些容器是从标准的Python容器(例如对象,列表和字典)继承的,以使它们可以与其他工具和库一起使用。

1.1 Basic containers

  • ObjectContainer

dcase_util.containers.ObjectContainer,从标准对象类继承的对象的容器类。

调用格式功能
ObjectContainer(\*args, \*\*kwargs)从标准对象类继承的对象的容器类。
ObjectContainer.load([filename])加载文件
ObjectContainer.save([filename])保存文件
ObjectContainer.show()打印容器内容
ObjectContainer.log([level])日志容器内容

示例如何从ObjectContainer继承类:

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
class Multiplier(dcase_util.containers.ObjectContainer):
def __init__(self, multiplier=None, **kwargs):
super(Multiplier, self).__init__(**kwargs)
self.multiplier = multiplier

def __str__(self):
ui = dcase_util.ui.FancyStringifier()
output = super(Multiplier, self).__str__()
output+= ui.data(field='multiplier', value=self.multiplier)
return output

def multiply(self, data):
return self.multiplier * data

m = Multiplier(10)
m.show()
# Multiplier :: Class
# multiplier : 10

print(m.multiply(5))
# 50

# Save object
m.save('test.cpickle')

# Load object
m2 = Multiplier().load('test.cpickle')

print(m2.multiply(5))
# 50

m2.show()
# Multiplier :: Class
# filename : test.cpickle
# multiplier : 10
1.1.1 DictContainer

dcase_util.containers.DictContainer字典容器类继承自标准的字典类。

调用格式功能
DictContainer(\*args, \*\*kwargs)字典从标准dict类继承的容器类。
DictContainer.load([filename])加载文件
DictContainer.save([filename])保存文件
DictContainer.show()打印容器内容
DictContainer.log([level])日志容器内容
DictContainer.get_path(path[, default, data])从带嵌入字典的路径中获取值
DictContainer.set_path(path, new_value[, data])使用点路径(dotted path)在嵌套字典中设置值
DictContainer.get_leaf_path_list([...])获取路径列表到嵌套字典中的所有叶节点。
DictContainer.merge(override[, target])递归字典合并
DictContainer.get_hash_for_path([dotted_path])在给定路径下获取数据的唯一哈希字符串。
DictContainer.get_hash([data])获取给定参数字典的唯一哈希字符串(md5)。

用法示例:

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
import dcase_util

d = dcase_util.containers.DictContainer(
{
'test': {
'field1': 1,
'field2': 2,
},
'test2': 100
}
)
d.show()
# DictContainer
# test
# field1 : 1
# field2 : 2
# test2 : 100

print(d.get_path('test.field1'))
# 1

print(d.get_path(['test', 'field1']))
# 1

print(d.get_path('test2'))
# 100

d.set_path('test.field2', 200)
print(d.get_path('test.field2'))
# 200

print(d.get_leaf_path_list())
# ['test.field1', 'test.field2', 'test2']

print(d.get_leaf_path_list(target_field_startswith='field'))
# ['test.field1', 'test.field2']

d.show()
# DictContainer
# test
# field1 : 1
# field2 : 200
# test2 : 100
1.1.2 ListContainer

dcase_util.containers.ListContainer,列表容器从标准列表类继承的容器类。

调用格式功能
ListContainer(\*args, \*\*kwargs)字典从标准dict类继承的容器类。
ListContainer.load([filename, headers])加载文件
ListContainer.save([filename])保存文件
ListContainer.show()打印容器内容
ListContainer.log([level])日志容器内容
ListContainer.update(data)用给定列表替换内容
1.1.3 ListDictContainer

dcase_util.containers.ListDictContainer,从标准列表类继承的字典列表容器类。

调用格式功能
ListDictContainer(\*args, \*\*kwargs)字典列表从标准dict类继承的容器类。
ListDictContainer.load([filename, fields, ...])加载文件
ListDictContainer.save([filename, fields, ...])保存文件
ListDictContainer.show()打印容器内容
ListDictContainer.log([level])日志容器内容
ListDictContainer.search(key, value)在字典列表中搜索
ListDictContainer.get_field(field_name[, ...])从字段获取所有数据

用法示例:

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
import dcase_util
ld = dcase_util.containers.ListDictContainer(
[
{
'field1': 1,
'field2': 2,
},
{
'field1': 3,
'field2': 4,
},
]
)
ld.show()
# ListDictContainer
# [0] =======================================
# DictContainer
# field1 : 1
# field2 : 2
#
# [1] =======================================
# DictContainer
# field1 : 3
# field2 : 4

print(ld.search(key='field1', value=3))
# DictContainer
# field1 : 3
# field2 : 4

print(ld.get_field(field_name='field2'))
# [2, 4]
1.1.4 RepositoryContainer

dcase_util.containers.RepositoryContainer,存储库的容器类,从dcase_util.containers.DictContainer继承。

调用格式功能
RepositoryContainer(\*args, \*\*kwargs)从DictContainer继承的存储库的容器类。
RepositoryContainer.load([filename])加载文件
RepositoryContainer.save([filename])保存文件
RepositoryContainer.show()打印容器内容
RepositoryContainer.log([level])日志容器内容
1.1.5 TextContainer

dcase_util.containers.TextContainer文本的容器类,从dcase_util.containers.ListContainer继承。

TextContainer(\ args,\ \ * kwargs)继承自ListContainer的文本的容器类。

调用格式功能
TextContainer(\*args, \*\*kwargs)继承自ListContainer的文本的容器类。
TextContainer.load([filename, headers])加载文件
TextContainer.save([filename])保存文件
TextContainer.show()打印容器内容
TextContainer.log([level])日志容器内容

1.2 Data containers

1.2.1 DataContainer

dcase_util.containers.DataContainer,数据的容器类,从dcase_util.containers.ObjectContainer继承。

调用格式功能
DataContainer([data, stats, metadata, ...])数据的容器类,从ObjectContainer继承
DataContainer.load([filename])Load file
DataContainer.save([filename])Save file
DataContainer.show()打印容器内容
DataContainer.log([level])记录容器内容
DataContainer.data数据矩阵
DataContainer.shape数据矩阵的形状
DataContainer.length数据列的数量
DataContainer.frames数据帧的数量
DataContainer.push_processing_chain_item(...)加工链项目入栈
DataContainer.focus_start聚焦段开始
DataContainer.focus_stop聚焦段结束
DataContainer.stats数据矩阵的基本统计
DataContainer.reset_focus()重置聚焦段(Focus segment)
DataContainer.get_focused()从数据数组获取聚焦段
DataContainer.freeze()冻结聚焦段(Freeze focus segment),并复制为容器数据。
DataContainer.get_frames([frame_ids, frame_hop])从数据数组中获取帧。
DataContainer.plot()可视化数据数组。
1.2.2 DataArrayContainer
`dcase_util.containers.DataArrayContainer`

Container class for data, inherited from dcase_util.containers.DataContainer.

调用格式功能
DataArrayContainer([data, stats, metadata, ...])Array data container DataArrayContainer.load([filename])Load file
DataArrayContainer.save([filename])Save file
DataArrayContainer.show(Print container content
DataArrayContainer.log([level])Log container content
DataArrayContainer.dataData matrix
DataArrayContainer.shapeShape of data matrix
DataArrayContainer.lengthNumber of data columns
DataArrayContainer.framesNumber of data frames
DataArrayContainer.push_processing_chain_item(...)Push processing chain item
ataArrayContainer.focus_startFocus segment start
ataArrayContainer.focus_stopFocus segment stop
DataArrayContainer.statsBasic statistics of data matrix.
DataArrayContainer.reset_focus()Reset focus segment
DataArrayContainer.get_focused()Get focus segment from data array.
DataArrayContainer.freeze()Freeze focus segment, copy segment to be container’s data.
DataArrayContainer.get_frames([frame_ids, ...])Get frames from data array.
DataArrayContainer.plot()Visualize data array.
1.2.3 DataMatrix2DContainer
`dcase_util.containers.DataMatrix2DContainer`

DataMatrix2DContainer是二维数据矩阵(numpy.ndarray)的数据容器。

基本用法:

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
# Initialize container with random matrix 10x100, and set time resolution to 20ms
data_container = dcase_util.containers.DataMatrix2DContainer(
data=numpy.random.rand(10,100),
time_resolution=0.02
)

# When storing, e.g., acoustic features, time resolution corresponds to feature extraction frame hop length.

# Access data matrix directly
print(data_container.data.shape)
# (10, 100)

# Show container information
data_container.show()
# DataMatrix2DContainer :: Class
# Data
# data : matrix (10,100)
# Dimensions
# time_axis : 1
# data_axis : 0
# Timing information
# time_resolution : 0.02 sec
# Meta
# stats : Calculated
# metadata : -
# processing_chain : -
# Duration
# Frames : 100
# Seconds : 2.00 sec

该容器具有聚焦机制,可灵活定位数据矩阵的一部分。 可以根据时间进行对焦(如果时间分辨率已定义,则以秒为单位),或基于帧ID。

使用焦点机制( focus mechanism),访问数据和可视化数据的示例:

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
# Using focus to get part data between 0.5 sec and 1.0 sec
print(data_container.set_focus(start_seconds=0.5, stop_seconds=1.0).get_focused().shape)
# (10, 25)

# Using focus to get part data between frame 10 and 50
print(data_container.set_focus(start=10, stop=50).get_focused().shape)
# (10, 40)

# Resetting focus and accessing full data matrix
data_container.reset_focus()
print(data_container.get_focused().shape)
# (10, 100)

# Access frames 1, 2, 10, and 30
data_container.get_frames(frame_ids=[1,2,10,30])

# Access frames 1-5, and only first value per column
data_container.get_frames(frame_ids=[1,2,3,4,5], vector_ids=[0])

# Transpose matrix
transposed_data = data_container.T
print(transposed_data.shape)
# (100, 10)

# Plot data
data_container.plot()
调用格式功能
DataMatrix2DContainer([data, stats, ...])Two-dimensional data matrix container class, inherited from DataContainer.
DataMatrix2DContainer.load([filename])Load file
DataMatrix2DContainer.save([filename])Save file
DataMatrix2DContainer.show()Print container content
DataMatrix2DContainer.log([level])Log container content
DataMatrix2DContainer.dataData matrix
DataMatrix2DContainer.shapeShape of data matrix
DataMatrix2DContainer.lengthNumber of data columns
DataMatrix2DContainer.framesNumber of data frames
DataMatrix2DContainer.vector_lengthData vector length
DataMatrix2DContainer.push_processing_chain_item(...)Push processing chain item
DataMatrix2DContainer.focus_startFocus segment start
DataMatrix2DContainer.focus_stopFocus segment stop
DataMatrix2DContainer.TTransposed data in a data container
DataMatrix2DContainer.statsBasic statistics of data matrix.
DataMatrix2DContainer.reset_focus()Reset focus segment
DataMatrix2DContainer.get_focused()Get focus segment from data matrix.
DataMatrix2DContainer.freeze()Freeze focus segment, copy segment to be container’s data.
DataMatrix2DContainer.get_frames([...])Get frames from data matrix.
DataMatrix2DContainer.plot()Visualize data matrix.
1.2.4 DataMatrix3DContainer
`dcase_util.containers.DataMatrix3DContainer`
调用格式功能
DataMatrix3DContainer([data, stats, ...])三位数据矩阵容器, 从DataMatrix2DContainer继承.
DataMatrix3DContainer.load([filename])Load file
DataMatrix3DContainer.save([filename])Save file
DataMatrix3DContainer.show()Print container content
DataMatrix3DContainer.log([level])`Log container content
DataMatrix3DContainer.dataData matrix
DataMatrix3DContainer.lengthNumber of data columns
DataMatrix3DContainer.framesNumber of data frames
1.2.5 BinaryMatrix3DContainer
`dcase_util.containers.BinaryMatrix2DContainer`
调用格式功能
BinaryMatrix2DContainer([data, ...])Two-dimensional data matrix container class, inherited from DataContainer.
BinaryMatrix2DContainer.load([filename])Load file
BinaryMatrix2DContainer.save([filename])Save file
BinaryMatrix2DContainer.show()Print container content
BinaryMatrix2DContainer.log([level])Log container content
BinaryMatrix2DContainer.dataData matrix
BinaryMatrix2DContainer.lengthNumber of data columns
BinaryMatrix2DContainer.framesNumber of data frames
BinaryMatrix2DContainer.pad(length[, ...])Pad binary matrix along time axis
BinaryMatrix2DContainer.plot([...])Visualize binary matrix, and optionally synced data matrix.
1.2.4 DataRepository
dcase_util.containers.DataRepository

DataRepository是可用于存储多个其他数据容器的容器。存储库存储具有两个级别信息的数据:标签和流。 标签是更高级别的密钥,流是第二级。 例如,可以使用储存库来储存与相同音频信号有关的多个不同声学特征。 流ID可用于存储从不同音频通道提取的特征。 后面的功能可以使用提取器标签和流ID进行访问。

用法示例:

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
# Initialize container with data
data_repository = dcase_util.containers.DataRepository(
data={
'label1': {
'stream0': {
'data': 100
},
'stream1': {
'data': 200
}
},
'label2': {
'stream0': {
'data': 300
},
'stream1': {
'data': 400
}
}
}
)
# Show container information::
data_repository. show()
# DataRepository :: Class
# Repository info
# Item class : DataMatrix2DContainer
# Item count : 2
# Labels : ['label1', 'label2']
# Content
# [label1][stream1] : {'data': 200}
# [label1][stream0] : {'data': 100}
# [label2][stream1] : {'data': 400}
# [label2][stream0] : {'data': 300}

# Accessing data inside repository
data_repository.get_container(label='label1',stream_id='stream1')
# {'data': 200}

# Setting data
data_repository.set_container(label='label3',stream_id='stream0', container={'data':500})
data_repository. show()
# DataRepository :: Class
# Repository info
# Item class : DataMatrix2DContainer
# Item count : 3
# Labels : ['label1', 'label2', 'label3']
# Content
# [label1][stream1] : {'data': 200}
# [label1][stream0] : {'data': 100}
# [label2][stream1] : {'data': 400}
# [label2][stream0] : {'data': 300}
# [label3][stream0] : {'data': 500}
调用格式功能
DataRepository([data, filename_dict, ...])数据存储库容器类将多个DataContainer一起存储
DataRepository.load([filename_dict])Load file list
DataRepository.get_container(label[, stream_id])Get container from repository
DataRepository.set_container(container, label)Store container to repository
DataRepository.push_processing_chain_item(...)Push processing chain item
DataRepository.plot()可视化存储在存储库中的数据。

1.3 Audio containers

dcase_util.containers.AudioContainer

AudioContainer是用于多声道音频的数据容器。 它读取多种格式(WAV,FLAC,M4A,WEBM)并写入WAV和FLAC文件。 直接从Youtube下载音频内容也受支持。

基本用法示例:

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
# Generating two-channel audio
audio_container = dcase_util.containers.AudioContainer(fs=44100)
t = numpy.linspace(0, 2, 2 * audio_container.fs, endpoint=False)
x1 = numpy.sin(220 * 2 * numpy.pi * t)
x2 = numpy.sin(440 * 2 * numpy.pi * t)
audio_container.data = numpy.vstack([x1, x2])

audio_container.show()
# AudioContainer :: Class
# Sampling rate : 44100
# Channels : 2
# Duration
# Seconds : 2.00 sec
# Milliseconds : 2000.00 ms
# Samples : 88200 samples

# Loading audio file
audio_container = dcase_util.containers.AudioContainer().load(
filename=dcase_util.utils.Example.audio_filename()
)

# Loading audio content from Youtube
audio_container = dcase_util.containers.AudioContainer().load_from_youtube(
query_id='2ceUOv8A3FE',
start=1,
stop=5
)

该容器具有聚焦机制,可灵活捕捉部分音频数据,同时保持完整的音频信号不变。 可以根据时间进行聚焦(如果时间分辨率已定义,则以秒为单位),或基于样本ID。 可以对单声道或混音(单声道)频道进行聚焦。 音频容器内容可以通过冻结来替代焦点细分。

使用焦点细分机制的示例:

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
# Using focus to get part data between 0.5 sec and 1.0 sec
print(audio_container.set_focus(start_seconds=0.5, stop_seconds=1.0).get_focused().shape)
# (2, 22050)

# Using focus to get part data starting 5 sec with 2 sec duration
print(audio_container.set_focus(start_seconds=5, duration_seconds=2.0).get_focused().shape)
# (2, 88200)

# Using focus to get part data starting 5 sec with 2 sec duration, mixdown of two stereo channels
print(audio_container.set_focus(start_seconds=5, duration_seconds=2.0, channel='mixdown').get_focused().shape)
# (88200,)

# Using focus to get part data starting 5 sec with 2 sec duration, left of two stereo channels
print(audio_container.set_focus(start_seconds=5, duration_seconds=2.0, channel='left').get_focused().shape)
# (88200,)

# Using focus to get part data starting 5 sec with 2 sec duration, seconds audio channel (indexing starts from 0)
print(audio_container.set_focus(start_seconds=5, duration_seconds=2.0, channel=1).get_focused().shape)
# (88200,)

# Using focus to get part data between samples 44100 and 88200
print(audio_container.set_focus(start=44100, stop=88200).get_focused().shape)
# (2, 44100)

# Resetting focus and accessing full data matrix::
audio_container.reset_focus()
print(audio_container.get_focused().shape)
# (2, 441001)

# Using focus to get part data starting 5 sec with 2 sec duration, and freeze this segment ::
audio_container.set_focus(start_seconds=5, duration_seconds=2.0).freeze()
print(audio_container.shape)
# (2, 88200)

Processing examples:

1
2
3
4
5
# Normalizing audio
audio_container.normalize()

# Resampling audio to target sampling rate
audio_container.resample(target_fs=16000)

Visualizations examples:

1
2
3
4
5
# Plotting waveform
audio_container.plot_wave()

# Plotting spectrogram
audio_container.plot_spec()
功能
AudioContainer([data, fs, ...])Audio container class.
AudioContainer.load([filename, fs, mono, ...])Load file
AudioContainer.load_from_youtube(query_id[, ...])Load audio data from youtube
AudioContainer.save([filename, bit_depth])`Save audio
AudioContainer.show()Print container content
AudioContainer.log([level])Log container content
AudioContainer.dataAudio data
AudioContainer.focus_start_samplesFocus segment start in samples.
AudioContainer.focus_start_secondsFocus segment start in seconds.
AudioContainer.focus_stop_samplesFocus segment stop in samples.
AudioContainer.focus_stop_secondsFocus segment stop in seconds.
AudioContainer.focus_channelFocus channel
AudioContainer.loadedAudio load status.
AudioContainer.shapeAudio data shape.
AudioContainer.lengthLength of audio data in samples.
AudioContainer.duration_samplesDuration of audio data in samples.
AudioContainer.duration_msDuration of audio data in milliseconds.
AudioContainer.duration_secDuration of audio data in seconds.
AudioContainer.streamsRename channels for compatibility.
AudioContainer.emptCheck if audio data is empty.
AudioContainer.reset_focus()Reset focus segment.
AudioContainer.set_focus([start, stop, ...])Set focus segment
AudioContainer.get_focused()Get focus segment from audio data.
AudioContainer.freeze()Freeze focus segment, copy segment to be container’s data.
AudioContainer.frames([frame_length, ...])Slice audio into overlapping frames.
AudioContainer.normalize([headroom])Normalize audio data.
AudioContainer.resample(target_fs[, scale, ...])Resample audio data.
AudioContainer.mixdown()Mix all audio channels into single channel.
AudioContainer.plot([plot_type])Visualize audio data
AudioContainer.plot_wave([x_axis, ...])Visualize audio data as waveform.
AudioContainer.plot_spec([spec_type, ...])Visualize audio data as spectrogram.

1.4 Feature containers

  • FeatureContainer

    dcase_util.containers.FeatureContainer

功能
FeatureContainer([data, stats, metadata, ...])从DataContainer继承的单个特征矩阵的特征容器类。
  • FeatureRepository

    dcase_util.containers.FeatureRepository

功能
FeatureRepository([filename_dict, ...])Feature repository container class to store multiple FeatureContainers together.

1.5 Mapping containers

OneToOneMappingContainerdcase_util.containers.OneToOneMappingContainer

功能
OneToOneMappingContainer(\*args, \*\*kwargs)Mapping container class for 1:1 data mapping, inherited from DictContainer class.
OneToOneMappingContainer.load([filename])Load file
OneToOneMappingContainer.save([filename])Save file
OneToOneMappingContainer.show()Print container content
OneToOneMappingContainer.log([level])Log container content
OneToOneMappingContainer.map(key[, default])Map with a key.
OneToOneMappingContainer.flippedExchange map key and value pairs.

1.6 Metadata containers

  • MetaDataItem

    dcase_util.containers.MetaDataItem

功能
MetaDataItem(\*args, \*\*kwargs)Meta data item class, inherited from standard dict class.
MetaDataItem.show()Print container content
MetaDataItem.log([level])Log container content
MetaDataItem.idUnique item identifier
MetaDataItem.get_list()Return item values in a list with specified order.
MetaDataItem.filenameFilename
MetaDataItem.filename_originalFilename
MetaDataItem.scene_labelScene label
MetaDataItem.event_labelEvent label
MetaDataItem.onsetOnset
MetaDataItem.offsetOffset
MetaDataItem.identifierIdentifier
MetaDataItem.source_labelSource label
MetaDataItem.tagsTags
MetaDataItem.active_within_segment(start, stop)Item active withing given segment.
  • MetaDataContainer

    dcase_util.containers.MetaDataContainer

功能
MetaDataContainer(\*args, \*\*kwargs)Meta data container class, inherited from ListDictContainer.
MetaDataContainer.log([level, show_data, ...])Log container content
MetaDataContainer.log_all([level])Log container content with all meta data items.
MetaDataContainer.show([show_data, show_stats])Print container content
MetaDataContainer.show_all()Print container content with all meta data items.
MetaDataContainer.load([filename, fields, ...])Load event list from delimited text file (csv-formatted)
MetaDataContainer.save([filename, fields, ...])Save content to csv file
MetaDataContainer.append(item)Append item to the meta data list
MetaDataContainer.file_countNumber of files
MetaDataContainer.event_countNumber of events
MetaDataContainer.scene_label_countNumber of unique scene labels
MetaDataContainer.event_label_countNumber of unique event labels
MetaDataContainer.identifier_countNumber of unique identifiers
MetaDataContainer.tag_countNumber of unique tags
MetaDataContainer.unique_filesUnique files
MetaDataContainer.unique_event_labelsUnique event labels
MetaDataContainer.unique_scene_labelsUnique scene labels
MetaDataContainer.unique_tagsUnique tags
MetaDataContainer.unique_identifiersUnique identifiers
MetaDataContainer.max_offsetFind the offset (end-time) of last event
MetaDataContainer.get_string([show_data, ...])Get content in string format
MetaDataContainer.filter([filename, ...])Filter content
MetaDataContainer.filter_time_segment([...])Filter time segment
MetaDataContainer.process_events([...])Process event content
MetaDataContainer.add_time(time)Add time offset to event onset and offset timestamps
MetaDataContainer.stats([event_label_list, ...])Statistics of the container content
MetaDataContainer.scene_stat_counts()Scene count statistics
MetaDataContainer.event_stat_counts()Event count statistics
MetaDataContainer.tag_stat_counts()Tag count statistics
MetaDataContainer.to_event_roll([...])Event roll
MetaDataContainer.intersection(second_metadata)Intersection of two meta containers
MetaDataContainer.intersection_report(...)Intersection report for two meta containers
MetaDataContainer.difference(second_metadata)Difference of two meta containers

1.7 Parameter containers

  • ParameterContainer

    dcase_util.containers.ParameterContainer

类名功能
ParameterContainer(\*args, \*\*kwargs)参数容器类,用于继承自DictContainer类的参数。
  • AppParameterContainer

    dcase_util.containers.AppParameterContainer

类名功能
AppParameterContainer([data, app_base, ...])应用程序参数的参数容器类,继承自ParameterContainer
AppParameterContainer.reset([field_labels, ...])
AppParameterContainer.process([...])Process parameters
AppParameterContainer.override(override)递归地覆盖容器内容。
AppParameterContainer.get_path_translated(path)用路径获取数据,路径可以包含将被翻译的字符串常量。
AppParameterContainer.set_path_translated(...)用路径设置数据,路径可以包含将被翻译的字符串常量。
  • DCASEAppParameterContainer

    dcase_util.containers.DCASEAppParameterContainer

类名功能
DCASEAppParameterContainer(\*args, \*\*kwargs)DCASE应用程序参数文件的参数容器类,从AppParameterContainer继承。
  • ParameterListContainer

    dcase_util.containers.ParameterListContainer

类名功能
ParameterListContainer(\*args, \*\*kwargs)参数列表容器,继承自ListDictContainer。

1.8 Probability containers

  • ProbabilityItem

    dcase_util.containers.ProbabilityItem

类名功能
ProbabilityItem(\*args, \*\*kwargs)概率数据项类,继承自标准字典类。
ProbabilityItem.show()Print container content
ProbabilityItem.log([level])Log container content
ProbabilityItem.filenameFilename
ProbabilityItem.labelLabel
ProbabilityItem.probabilityReturns:
ProbabilityItem.id唯一的项目(item)标识
ProbabilityItem.get_list()以指定顺序返回列表中的项目(item)值。
  • ProbabilityContainer

    dcase_util.containers.ProbabilityContainer

类名功能
ProbabilityContainer(\*args, \*\*kwargs)概率数据容器类,继承自ListDictContainer。
ProbabilityContainer.show()Print container content
ProbabilityContainer.log([level])Log container content
ProbabilityContainer.load([filename])来自分隔文本文件的加载概率列表(csv格式)
ProbabilityContainer.save([filename, delimiter])Save content to csv file
ProbabilityContainer.append(item)Append item to the meta data list
ProbabilityContainer.unique_filesUnique files
ProbabilityContainer.unique_labelsUnique labels
ProbabilityContainer.filter([filename, ...])Filter content

1.9 MIxins

  • ContainerMixin

    dcase_util.containers.ContainerMixin

类名功能
ContainerMixin(\*args, \*\*kwargs)Container mixin to give class basic container methods.
ContainerMixin.show()Print container content
ContainerMixin.log([level])Log container content
  • FileMixin

    dcase_util.containers.FileMixin

类名功能
FileMixin(\*args, \*\*kwargs)File mixin to give class methods to load and store content.
FileMixin.get_file_information()Get file information, filename
FileMixin.detect_file_format([filename])Detect file format from extension
FileMixin.validate_format()Validate file format
FileMixin.exists()Checks that file exists
FileMixin.empty()Check if file is empty
FileMixin.delimiter([exclude_delimiters])Use csv.sniffer to guess delimiter for CSV file
FileMixin.is_package([filename])Determine if the file is compressed package.
  • PackageMixin

    dcase_util.containers.PackageMixin

类名功能
PackageMixin(\*args, \*\*kwargs)打包mixin以提供处理压缩文件包的类基本方法。
PackageMixin.package_passwordPackage password
PackageMixin.extract([overwrite, ...])解压缩包

2. Data

数据处理的类

2.1 Buffers

dcase_util.data.DataBuffer

数据缓冲类,可用于存储与项目关联的数据和元数据。 项目数据通过项目键进行访问。当内部缓冲区被填满时,最旧的项目被替换。

类名功能
DataBuffer([size])数据缓冲区(先进先出)
DataBuffer.set(key[, data, meta])将项目(item)插入缓冲区
DataBuffer.get(key)根据键获取项目(item)
DataBuffer.clear()清空缓冲区
DataBuffer.count缓冲区使用情况
DataBuffer.full缓冲区已满
DataBuffer.key_exists(key)检查键(key)是否存在于缓冲区中

2.2 Encoders

  • BinaryMatrixEncoder

    dcase_util.data.BinaryMatrixEncoder

类名功能
BinaryMatrixEncoder([label_list,...])二进制矩阵编码器基类
BinaryMatrixEncoder.pad(length [,binary_matrix])沿时间轴填充二进制矩阵
BinaryMatrixEncoder.plot([binary_matrix,...])可视化二进制矩阵和可选的同步数据矩阵。
  • OneHotEncoder

    dcase_util.data.OneHotEncoder

类名功能
OneHotEncoder([label_list, time_resolution, ...])One hot encoder class
OneHotEncoder.encode(label[, length_frames, ...])Generate one hot binary matrix
  • ManyHotEncoder

    dcase_util.data.ManyHotEncoder

类名功能
ManyHotEncoder([label_list, ...])Many hot encoder class
ManyHotEncoder.encode(label_list[, ...])Generate one hot binary matrix
  • EventRollEncoder

    dcase_util.data.EventRollEncoder

类名功能
EventRollEncoder([label_list, ...])Event list encoder class
EventRollEncoder.encode(metadata_container)Generate event roll from MetaDataContainer

2.3 Data manipulators

  • Normalizer

    dcase_util.data.Normalizer

类名功能
Normalizer([n, s1, s2, mean, std])数据归一化器来累积数据统计
Normalizer.log([level])记录容器内容
Normalizer.show()打印容器内容
Normalizer.load([filename])Load file
Normalizer.save([filename])Save file
Normalizer.meanMean vector
Normalizer.std标准差矢量
Normalizer.reset()充值内部变量
Normalizer.accumulate(data[, time_axis])计算统计量
Normalizer.finalize()完成统计计算
Normalizer.normalize(data)使用该类的内部统计量标准化特征矩阵
  • RepositoryNormalizer

    dcase_util.data.RepositoryNormalizer

类名功能
RepositoryNormalizer([normalizer_dict, ...])Data repository normalizer
RepositoryNormalizer.load(filename_dict)Load normalizers from disk.
RepositoryNormalizer.normalize(data_repository)Normalize data repository
  • Aggregator(聚合)
    dcase_util.data.Aggregator

数据聚合器可用于处理窗口中的数据矩阵。 这个处理阶段可以用来通过计算它们的平均值和标准差来在特定的窗口长度内折叠数据,或者将该矩阵平坦化为单个向量。

支持的处理方法:

1
2
3
4
5
6
7
8
9
10
data_aggregator = dcase_util.data.Aggregator(
recipe=['mean', 'std'],
win_length_frames=10,
hop_length_frames=1,
)

data_stacker = dcase_util.data.Stacker(recipe='mfcc')
data_repository = dcase_util.utils.Example.feature_repository()
data_matrix = data_stacker.stack(data_repository)
data_matrix = data_aggregator.aggregate(data_matrix)
类名功能
Aggregator([win_length_frames, ...])Data aggregator
Aggregator.log([level])Log container content
Aggregator.show()Print container content
Aggregator.load([filename])Load file
Aggregator.save([filename])Save file
Aggregator.aggregate([data])Aggregate data
  • Sequencer

    dcase_util.data.Sequencer

    Sequencer类将数据矩阵处理成序列(图像)。 序列可以重叠,并且可以在调用之间改变排序网格(移位)。

    类名 | 功能
    —|—
    Sequencer([frames, hop_length_frames, ...]) | Data sequencer
    Sequencer.log([level]) | Log container content
    Sequencer.show() | Print container content
    Sequencer.load([filename]) | Load file
    Sequencer.save([filename]) | Save file
    Sequencer.sequence([data]) | Make sequences
    Sequencer.increase_shifting([shift_step]) | Increase temporal shifting

  • Stacker

    dcase_util.data.Stacker

数据堆叠类。 Class使用矢量配方和DataRepository,并创建适当的数据矩阵。

矢量方法(Vector recipe)

使用recipe,您可以选择全矩阵,只选择开始和结束索引的一部分,或从中选择单个行。

例子:

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
[
{
'method': 'mfcc',
},
{
'method': 'mfcc_delta'
'vector-index: {
'channel': 0,
'start': 1,
'end': 17,
'full': False,
'selection': False,
}
},
{
'method': 'mfcc_acceleration',
'vector-index: {
'channel': 0,
'full': False,
'selection': True,
'vector': [2, 4, 6]
}
}
]

请参阅dcase_util.utils.VectorRecipeParser如何方便地使用配方字符串来生成上述数据结构。

类名功能
Stacker([recipe, hop])Data stacker
Stacker.log([level])Log container content
Stacker.show()Print container content
Stacker.load([filename])Load file
Stacker.save([filename])Save file
Stacker.stack(repository)Vector creation based on recipe
  • Selector

    dcase_util.data.Selector,数据选择类

类名功能
Selector(\*\*kwargs)Data selector
Selector.log([level])Log container content
Selector.show()Print container content
Selector.load([filename])Load file
Selector.save([filename])Save file
Selector.select(data[, selection_events])Selecting feature repository with given events
  • Masker

    dcase_util.data.Masker,数据屏蔽(筛选)类。

类名功能
Masker(\*\*kwargs)Data masker
Masker.log([level])Log container content
Masker.show()Print container content
Masker.load([filename])Load file
Masker.save([filename])Save file
Masker.mask(data[, mask_events])Masking feature repository with given events

2.4 Probability

ProbabilityEncoderdcase_util.data.ProbabilityEncoder

类名功能
ProbabilityEncoder([label_list])Constructor
ProbabilityEncoder.log([level])Log container content
ProbabilityEncoder.show()Print container content
ProbabilityEncoder.load([filename])Load file
ProbabilityEncoder.save([filename])Save file
ProbabilityEncoder.collapse_probabilities(...)Collapse probabilities
ProbabilityEncoder.collapse_probabilities_windowed(...)Collapse probabilities in windows
ProbabilityEncoder.binarization(probabilities)Binarization

2.5 Eecisions

DecisionEncoderdcase_util.data.DecisionEncoder

类名功能
DecisionEncoder([label_list])构造器
DecisionEncoder.log([level])Log container content
DecisionEncoder.show()Print container content
DecisionEncoder.load([filename])Load file
DecisionEncoder.save([filename])Save file
DecisionEncoder.majority_vote(frame_decisions)多票制
DecisionEncoder.find_contiguous_regions(...)从具有布尔值的numpy.array找到连续区域。
DecisionEncoder.process_activity(...[, operator])处理动态数组(二进制)

3. Datasets

3.1 Dataset

dcase_util.datasets.Dataset这是基类,所有专用数据集都是从它继承而来的。基类本身不能使用。

用法示例:

1
2
3
4
5
6
7
8
9
10
11
12
# Create class
dataset = dcase_util.datasets.TUTAcousticScenes_2017_DevelopmentSet(data_path='data')
# Initialize dataset, this will make sure dataset is downloaded, packages are extracted,
# and needed meta files are created
dataset.initialize()
# Show meta data
dataset.meta.show()
# Get all evaluation setup folds
folds = dataset.folds()
# Get all evaluation setup folds
train_data_fold1 = dataset.train(fold=folds[0])
test_data_fold1 = dataset.test(fold=folds[0])
语法格式功能
Dataset([name, storage_name, data_path, ...])数据集基类
Dataset.initialize()数据集基类
Dataset.download_packages()通过互联网将数据集包下载到本地路径
Dataset.extract_packages()提取数据集包
Dataset.prepare()为使用准备数据集
Dataset.process_meta_item(item[, absolute_path])处理单个元数据项目
Dataset.check_filelist()从文件列表生成哈希,并检查它是否与保存在filelist.hash中的哈希匹配
Dataset.show()打印数据集信息
Dataset.log()记录数据集信息
Dataset.load()将数据集元数据和交叉验证集加载到容器中
Dataset.load_meta()将元数据添加到容器中
Dataset.load_crossvalidation_data()将交叉验证加载到容器中
Dataset.audio_files获取数据集中的所有音频文件
Dataset.audio_file_count获取数据集中音频文件的数量
Dataset.meta获取数据集的元数据。
Dataset.meta_count元数据项的数量
Dataset.error_meta获取数据集的音频错误元数据
Dataset.error_meta_count错误元数据项的数量
Dataset.folds([mode])fold ID列表
Dataset.fold_count评估设置中的fold次数
Dataset.evaluation_setup_filename([...])评估设置文件名生成
Dataset.train([fold, absolute_paths])训练列表
Dataset.test([fold, absolute_paths])测试列表
Dataset.eval([fold, absolute_paths])评估列表
Dataset.train_files([fold, absolute_paths])训练文件列表
Dataset.test_files([fold, absolute_paths])测试文件列表
Dataset.eval_files([fold, absolute_paths])评估文件列表
Dataset.validation_split([fold, split_type, ...])验证文件列表
Dataset.validation_files_dataset([fold, verbose])由数据集提供的验证文件列表
Dataset.validation_files_random([fold, ...])从训练集中随机选择的验证文件列表。
Dataset.validation_files_balanced([fold, ...])在保持数据平衡的同时随机选择验证文件列表
Dataset.scene_labels()元数据中唯一场景标签的列表
Dataset.scene_label_count()元数据中唯一场景标签的数量
Dataset.event_labels(\*\*kwargs)元数据中的唯一事件标签列表
Dataset.event_label_count(\*\*kwargs)元数据中的唯一事件标签列表
Dataset.tags()元数据中唯一音频标签的列表
Dataset.tag_count()元数据中唯一音频标签的数量
Dataset.file_meta(filename)给定文件的元数据
Dataset.file_error_meta(filename)给定文件的错误元数据
Dataset.file_features(filename)预先计算给定文件的声学特征
Dataset.relative_to_absolute_path(path)将相对路径转换为绝对路径
Dataset.absolute_to_relative_path(path)将绝对路径转换为相对路径
Dataset.dataset_bytes()数据集的总下载大小(以字节为单位)
Dataset.dataset_size_string()字符串中数据集的总下载大小
Dataset.dataset_size_on_disk()当前存储在本地的数据集的总大小

3.2 AcousticScenesDataset

dcase_util.datasets.AcousticSceneDataset,格式如下:

AcousticSceneDataset(\*args, \*\*kwargs)

继承自AcousticSceneDataset的几个类:

类名功能
TUTAcousticScenes_2017_DevelopmentSet([...])TUT Acoustic scenes 2017 开发数据集
TUTAcousticScenes_2017_EvaluationSet([...])TUT Acoustic scenes 2017 评估数据集
TUTAcousticScenes_2016_DevelopmentSet([...])TUT Acoustic scenes 2016 开发数据集
TUTAcousticScenes_2016_EvaluationSet([...])TUT Acoustic scenes 2016 评估数据集

3.3 SoundEventDataset

dcase_util.datasets.SoundEventDataset

方法功能
SoundEventDataset(\*args, \*\*kwargs)
SoundEventDataset.event_label_count([...])Number of unique scene labels in the meta data.
SoundEventDataset.event_labels([scene_label])List of unique event labels in the meta data.
SoundEventDataset.train([fold, ...])List of training items.
SoundEventDataset.test([fold, ...])List of testing items.

继承自SoundEventDataset的几个类:

类名功能
TUTRareSoundEvents_2017_DevelopmentSet([...])TUT Acoustic scenes 2017 development dataset
TUTRareSoundEvents_2017_EvaluationSet([...])TUT Acoustic scenes 2017 evaluation dataset
TUTSoundEvents_2017_DevelopmentSet([...])TUT Sound events 2017 development dataset
TUTSoundEvents_2017_EvaluationSet([...])TUT Sound events 2017 evaluation dataset
TUTSoundEvents_2016_DevelopmentSet([...])TUT Sound events 2016 development dataset
TUTSoundEvents_2016_EvaluationSet([...])TUT Sound events 2016 evaluation dataset
TUT_SED_Synthetic_2016([storage_name, ...])TUT SED Synthetic 2016

3.4 AudioTaggingDataset

dcase_util.datasets.AudioTaggingDataset

语法格式:AudioTaggingDataset(\*args, \*\*kwargs)

类名功能
DCASE2017_Task4tagging_DevelopmentSet([...])DCASE 2017 Large-scale weakly supervised sound event detection for smart cars
DCASE2017_Task4tagging_EvaluationSet([...])DCASE 2017 Large-scale weakly supervised sound event detection for smart cars
CHiMEHome_DomesticAudioTag_DevelopmentSet([...])Constructor

4. Decorators(修饰符)

用于修饰函数的辅助类

RunOnce

dcase_util.decorators.RunOnce

RunOnce(f) Decorator 类只允许执行一次

5. Features

提取特征的类

  • FeatureExtractor(基本特征提取)

dcase_util.features.FeatureExtractor,特征提取基本类,语法格式:FeatureExtractor([fs, win_length_samples, ...])

1
class dcase_util.features.FeatureExtractor(fs=44100, win_length_samples=None, hop_length_samples=None, win_length_seconds=0.04, hop_length_seconds=0.02, **kwargs)
  • SpectralFeatureExtractor
    dcase_util.features.SpectralFeatureExtractor
类名功能
SpectralFeatureExtractor([spectrogram_type, ...])特殊特征提取基础类
SpectralFeatureExtractor.get_window_function(n)窗函数
SpectralFeatureExtractor.get_spectrogram(y)谱图
  • MelExtractor(Mel)
    dcase_util.features.MelExtractor
类名功能
MelExtractor([fs, win_length_samples, ...])梅尔带能量特征提取类
MelExtractor.extract(y)提取音频信号的特征
  • MfccStaticExtractor(MFCC)
    dcase_util.features.MfccStaticExtractor
类名功能
MfccStaticExtractor([fs, ...])用于提取静态MFCC功能的特征提取器类
MfccStaticExtractor.extract(y)提取音频信号的特征
  • MfccDeltaExtractor(MFCC一阶导)

dcase_util.features.MfccDeltaExtractor

类名功能
MfccDeltaExtractor([fs, win_length_samples, ...])MFCC一阶导
MfccDeltaExtractor.extract(y)提取音频信号的特征
  • MfccAccelerationExtractor(MFCC二阶导)
    dcase_util.features.MfccAccelerationExtractor
类名功能
MfccAccelerationExtractor([fs, ...])(MFCC二阶导 MFCC acceleration features
MfccAccelerationExtractor.extract(y)提取音频信号的特征
  • ZeroCrossingRateExtractor(过零率)
    dcase_util.features.ZeroCrossingRateExtractor
类名功能
ZeroCrossingRateExtractor([fs, ...])过零率
ZeroCrossingRateExtractor.extract(y)提取音频信号的特征
  • RMSEnergyExtractor(均方根能量特征)
    dcase_util.features.RMSEnergyExtractor
类名功能
RMSEnergyExtractor([fs, win_length_samples, ...])均方根能量特征
RMSEnergyExtractor.extract(y)提取音频信号的特征
  • SpectralCentroidExtractor(光谱质心)
    dcase_util.features.SpectralCentroidExtractor
类名功能
SpectralCentroidExtractor([fs, ...])光谱质心
SpectralCentroidExtractor.extract(y)提取音频信号的特征

6. Files

6.1 File

dcase_util.files.File,通用的文件类:

类名功能
File(\*args, \*\*kwargs)通用类
File.load([filename])加载文件
File.save(data[, filename])保存文件
File.get_file_information()得到文件信息、文件名
File.detect_file_format([filename])根据扩展检测文件格式
File.validate_format()验证文件格式有效性
File.exists()检查文件是否存在
File.empty()检查文件是否为空
File.delimiter([exclude_delimiters])使用csv.sniffer猜测CSV文件的分隔符
File.is_package([filename])确定文件是否为压缩包

6.2 FileLock

dcase_util.files.FileLock,简单的基于文件的锁定类。

类名功能
FileLock(filename[, timeout, ...])简单的基于文件的锁定类
FileLock.lock()锁定文件
FileLock.release()释放文件锁
FileLock.expired检查比指定超时更早的锁定文件
FileLock.is_locked检查锁定文件是否存在
FileLock.touch()使用当前时间戳创建锁定文件

6.3 remotefile

dcase_util.files.RemoteFile远程文件处理类。

类名功能
RemoteFile([filename,content_type,...])远程文件类
RemoteFile.download()下载远程文件并将其保存为本地文件。
RemoteFile.is_content_type(content_type)检查该文件是否包含给定类型的内容
RemoteFile.local_md5本地文件的校验和。
RemoteFile.local_modified修改本地文件的时间戳。
RemoteFile.local_bytes本地文件的文件大小(以字节为单位)。
RemoteFile.local_size_string()以可读形式存在的本地文件的文件大小。
RemoteFile.local_exists()检查本地文件是否存在。
RemoteFile.local_changed()检查本地文件是否对应于远程文件(基于校验和或修改时间和文件大小)。
RemoteFile.remote_file指向远程文件的URL
RemoteFile.remote_modified远程文件的最后修改时间。
RemoteFile.remote_bytes远程文件的文件大小。
RemoteFile.remote_status远程文件的状态。
RemoteFile.remote_size_string()以人类可读形式的远程文件的文件大小。
RemoteFile.remote_info()获取有关删除文件(状态,大小,校验和,上次修改时间)的信息。
RemoteFile.remote_exists()检查远程文件是否存在(基于HTTP状态码)。

6.4 RemotePackage

dcase_util.files.RemotePackage,远程包处理类。

类名功能
RemotePackage([filename,content_type,...])远程软件包类
RemotePackage.download()下载远程文件并将其保存为本地文件。
RemotePackage.extract([覆盖,...])解压缩包
RemotePackage.package_password包密码
RemotePackage.is_content_type(content_type)检查该文件是否包含给定类型的内容
RemotePackage.local_md5本地文件的校验和。
RemotePackage.local_modified修改本地文件的时间戳。
RemotePackage.local_bytes本地文件的文件大小(以字节为单位)。
RemotePackage.local_size_string()以可读形式存储本地文件的文件大小。
RemotePackage.local_exists()检查本地文件是否存在。
RemotePackage.local_changed()检查本地文件是否对应于远程文件(基于校验和或修改时间和文件大小)。
RemotePackage.remote_file指向远程文件的URL
RemotePackage.remote_modified远程文件的最后修改时间。
RemotePackage.remote_bytes远程文件的文件大小。
RemotePackage.remote_status远程文件的状态。
RemotePackage.remote_size_string()以可读形式显示远程文件的文件大小。
RemotePackage.remote_info()获取有关删除文件(状态,大小,校验和,上次修改时间)的信息。
RemotePackage.remote_exists()检查远程文件是否存在(基于HTTP状态码)。

6.5 Serializer

dcase_utils.files.Serializer,数据序列化类

类名功能
Serializer数据序列化类
Serializer.load_yaml(filename)加载YAML文件
Serializer.load_cpickle(filename)加载CPICKLE文件
Serializer.load_json(filename)加载JSON文件
Serializer.load_msgpack(filename)Load MSGPACK file
Serializer.load_marshal(filename)Load MARSHAL file
Serializer.save_yaml(filename, data)Save data into YAML file
Serializer.save_cpickle(filename, data)Save data into CPICKLE file
Serializer.save_json(filename, data)Save data into JSON file
Serializer.save_msgpack(filename, data)Save data into MSGPACK file
Serializer.save_marshal(filename, data)Save data into MARSHAL file

7. Keras utilities

使用Keras深度学习库的单元。

7.1 Model

dcase_util.keras.model。*

类名功能
create_sequential_model(model_parameter_list)创建顺序Keras模型
model_summary_string(keras_model)字符串中的模型摘要,类似于Keras模型摘要函数。

7.2 Callbacks

用法示例如何将外部度量与dcase_util提供的Callback类一起使用:

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
epochs = 100
batch_size = 256
loss = 'categorical_crossentropy'
metrics = ['categorical_accuracy']
processing_interval = 1
manual_update = True
external_metric_labels={'ER': 'Error rate'}

callback_list = [
dcase_util.keras.ProgressLoggerCallback(
epochs=epochs,
metric=metrics,
loss=loss,
manual_update=manual_update,
manual_update_interval=processing_interval,
external_metric_labels=external_metric_labels
),
dcase_util.keras.ProgressPlotterCallback(
epochs=epochs,
metric=metrics,
save=False,
manual_update=manual_update,
manual_update_interval=processing_interval,
external_metric_labels=external_metric_labels
),
dcase_util.keras.StopperCallback(
epochs=epochs,
monitor=metric[0],
manual_update=manual_update,
),
dcase_util.keras.StasherCallback(
epochs=epochs,
monitor=metric[0],
manual_update=manual_update,
)
]

for epoch_start in range(0, epochs, processing_interval):
epoch_end = epoch_start + processing_interval

# Make sure we have only specified amount of epochs
if epoch_end > epochs:
epoch_end = epochs

# Train model
keras_model.fit(
x=training_X,
y=training_Y,
validation_data=(validation_X, validation_Y),
callbacks=callback_list,
verbose=0,
initial_epoch=epoch_start,
epochs=epoch_end,
batch_size=batch_size,
shuffle=True
)
# Calculate external metrics
ER = 0.0

# Inject external metric values to the callbacks
for callback in callback_list:
if hasattr(callback, 'set_external_metric_value'):
callback.set_external_metric_value(
metric_label='ER',
metric_value=ER
)

# Manually update callbacks
for callback in callback_list:
if hasattr(callback, 'update'):
callback.update()

# Check we need to stop training
stop_training = False
for callback in callback_list:
if hasattr(callback, 'stop'):
if callback.stop():
stop_training = True

if stop_training:
# Stop the training loop
break
  • ProgressLoggerCallback
    dcase_util.keras.ProgressLoggerCallback,Keras回调用于存储tqdm进度条或日志记录界面的指标。 实现Keras回调API。

    此回调与标准的ProgbarLogger Keras回调非常相似,但它增加了对日志接口和外部度量(在Keras训练过程之外计算的度量)的支持。

类名功能
ProgressLoggerCallback([manual_update,...])Keras回调在日志界面中显示指标。
  • ProgressPlotterCallback
    dcase_util.keras.ProgressPlotterCallback,keras回调在培训过程中计划进度并将最终进展保存到数字中。 实现Keras回调API。
类名功能
ProgressPlotterCallback([epochs,...])Keras回调在训练过程中绘制进度并将最终进度保存到图中。
  • StopperCallback
    dcase_util.keras.StopperCallback,keras回调停止训练时,改善没有在规定数量的时代看到。 实现Keras回调API。

    此回调与标准的EarlyStopping Keras回调非常相似,但它增加了对外部度量标准(在Keras培训过程之外计算的度量)的支持。

类名功能
StopperCallback([epochs,manual_update,...])Keras回调在特定时间段内没有发现改进时停止训练。
  • StasherCallback

dcase_util.keras.StasherCallback,keras回调监测训练过程并存储最佳模型。实现Keras回调API。

该回调与标准的ModelCheckpoint Keras回调非常相似,但它增加了对外部度量(在Keras培训过程之外计算的度量)的支持。

类名功能
StasherCallback([epochs,manual_update,...])Keras回调监视训练过程并存储最佳模型。
  • BaseCallback

dcase_util.keras.BaseCallback

类名功能
BaseCallback([epochs, manual_update, ...])回调基础类

7.3 Utils

dcase_util.keras.utils. *

类名功能
setup_keras仅执行一次的修饰类

8. Processors(数据处理器类)

8.1 Processing chain

  • ProcessingChainItem
    dcase_util.processors.ProcessingChainItem
类名功能
ProcessingChainItem(\*args, \*\*kwargs)
  • ProcessingChain
    dcase_util.processors.ProcessingChain
类名功能
ProcessingChain(\*args, \*\*kwargs)
ProcessingChain.show_chain()显示链信息
ProcessingChain.log_chain([level])记录链信息
ProcessingChain.push_processor(processor_name)将处理器项目推送到链中。
ProcessingChain.process([data])用处理链处理数据
ProcessingChain.call_method(method_name[, ...])调用处理链项目中的类方法
ProcessingChain.processor_exists(processor_name)检查处理器是否存在于链中
ProcessingChain.processor_class_reference(...)处理器类的引用
ProcessingChain.processor_class(...)初始化处理器类

8.2 Audio

  • AudioReadingProcessor
    dcase_util.processors.AudioReadingProcessor
类名功能
AudioReadingProcessor([data, fs, ...])构造函数
AudioReadingProcessor.process([data, ...])音频读取
  • MonoAudioReadingProcessor
    dcase_util.processors.MonoAudioReadingProcessor
类名功能
MonoAudioReadingProcessor([data, fs, …])构造函数
MonoAudioReadingProcessor.process([data, …])音频读取

8.3 Data

  • AggregationProcessor
    dcase_util.processors.AggregationProcessor
类名功能
AggregationProcessor([win_length_frames, ...])Data aggregation processor
AggregationProcessor.process([data])Process features
  • SequencingProcessor
    dcase_util.processors.SequencingProcessor
类名功能
SequencingProcessor([frames, ...])Data sequencing processor
SequencingProcessor.process([data])Process
  • NormalizationProcessor
    dcase_util.processors.NormalizationProcessor
类名功能
NormalizationProcessor([n, s1, s2, mean, std])Data normalizer to accumulate data statistics
NormalizationProcessor.process([data])Normalize feature matrix with internal statistics of the class
  • RepositoryNormalizationProcessor
    dcase_util.processors.RepositoryNormalizationProcessor
类名功能
RepositoryNormalizationProcessor([parameters])Data normalizer to accumulate data statistics inside repository
RepositoryNormalizationProcessor.process([data])Normalize data repository with internal statistics
  • StackingProcessor
    dcase_util.processors.StackingProcessor
类名功能
StackingProcessor([recipe, hop])Data stacking processor
StackingProcessor.process([data])Vector creation based on recipe
  • OneHotEncodingProcessor
    dcase_util.processors.OneHotEncodingProcessor
类名功能
OneHotEncodingProcessor([label_list, ...])Event roll encoding processor
OneHotEncodingProcessor.process([data, ...])Encode metadata
  • ManyHotEncodingProcessor
    dcase_util.processors.ManyHotEncodingProcessor
类名功能
ManyHotEncodingProcessor([label_list, ...])Event roll encoding processor
ManyHotEncodingProcessor.process([data, ...])Encode metadata
  • EventRollEncodingProcessor
    dcase_util.processors.EventRollEncodingProcessor
类名功能
EventRollEncodingProcessor([label_list, ...])Event roll encoding processor
EventRollEncodingProcessor.process([data])Encode metadata

8.4 Features

  • RepositoryFeatureExtractorProcessor
    dcase_util.processors.RepositoryFeatureExtractorProcessor
类名功能
RepositoryFeatureExtractorProcessor([parameters])构造器
RepositoryFeatureExtractorProcessor.process([data])提取特征
  • FeatureExtractorProcessor
    dcase_util.processors.FeatureExtractorProcessor
类名功能
FeatureExtractorProcessor(\*args, \*\*kwargs)构造器
FeatureExtractorProcessor.process([data])提取特征
  • MelExtractorProcessor(Mel能量)
    dcase_util.processors.MelExtractorProcessor
类名功能
MelExtractorProcessor([fs, ...])构造器
MelExtractorProcessor.process([data])提取特征
  • MfccStaticExtractorProcessor(MFCC)
    dcase_util.processors.MfccStaticExtractorProcessor
类名功能
MfccStaticExtractorProcessor([fs, ...])构造器
MfccStaticExtractorProcessor.process([data])提取特征
  • MfccDeltaExtractorProcessor(MFCC一阶导)
    dcase_util.processors.MfccDeltaExtractorProcessor
类名功能
MfccDeltaExtractorProcessor([fs, ...])构造器
MfccDeltaExtractorProcessor.process([data])提取特征
  • MfccAccelerationExtractorProcessor(MFCC二阶导)
    dcase_util.processors.MfccAccelerationExtractorProcessor
类名功能
MfccAccelerationExtractorProcessor([fs, ...])构造器
MfccAccelerationExtractorProcessor.process([data])提取特征
  • ZeroCrossingRateExtractorProcessor(过零率)
    dcase_util.processors.ZeroCrossingRateExtractorProcessor
类名功能
ZeroCrossingRateExtractorProcessor([fs, ...])构造器
ZeroCrossingRateExtractorProcessor.process([data])提取特征
  • RMSEnergyExtractorProcessor(均方根能量)
    dcase_util.processors.RMSEnergyExtractorProcessor
类名功能
RMSEnergyExtractorProcessor([fs, ...])构造器
RMSEnergyExtractorProcessor.process([data])提取特征
  • SpectralCentroidExtractorProcessor(光谱质心)
    dcase_util.processors.SpectralCentroidExtractorProcessor
类名功能
SpectralCentroidExtractorProcessor([fs, ...])构造器
SpectralCentroidExtractorProcessor.process([data])提取特征

8.5 Metadata

MetadataReadingProcessordcase_util.processors.MetadataReadingProcessor

类名功能
MetadataReadingProcessor(\*args, \*\*kwargs)构造器
MetadataReadingProcessor.process([data, ...])读取元数据

8.6 Mixin

ProcessorMixindcase_util.processors.ProcessorMixin

类名功能
ProcessorMixin(\*args, \*\*kwargs)数据处理链单元混合
ProcessorMixin.process([data])处理数据
ProcessorMixin.get_processing_chain_item()使用当前处理器数据获取处理链项目
ProcessorMixin.push_processing_chain_item(...)推送加工链项目

9. User interfacing

用于轻型用户界面的实用程序类。

FancyLogger和FancyPrinter提供相同的API,唯一不同的是FancyLogger将输出到日志系统,FancyPrinter使用标准打印功能将输出打印到sys.stdout。 当需要以字符串形式输出时,FancyStringifier可用于一般情况。

9.1 FancyLogger

dcase_util.ui.FancyLogger

使用日志记录时,此类提供额外的格式。 如果在调用FancyLogger时Python日志记录尚未初始化,则首先调用dcase_util.utils.setup_logging

用法示例:

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
ui = dcase_util.ui.FancyLogger()
ui.title('title')
ui.section_header('section_header')
ui.sub_header('sub_header')
ui.foot('foot')
ui.line('line', indent=2)
ui.line('line', indent=4)
ui.line('line', indent=6)

# Data row with field and value
ui.data('data field', 'value', 'unit')

# Horizontal separator
ui.sep()

# Table
ui.table(cell_data=[[1, 2, 3], [1, 2, 3]])

# Faster way to create output tables without collecting data into one data structure.
ui.row('Header1', 'Header2', widths=[10,20], types=['float2','str20'])
ui.row('-','-')
ui.row(10.21231, 'String text')

输出:

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
[I] title
[I] section_header
[I] ========================================
[I] === sub_header ===
[I] foot
[I]
[I] line
[I] line
[I] line
[I] data field : value unit
[I] ========================================
[I] Col #0 Col #1
[I] ------ ------
[I] 1 1
[I] 2 2
[I] 3 3
[I]
[I] Header1 | Header2 |
[I] ------- | ----------------- |
[I] 10.21 | String text |
类名功能
FancyLogger()Logger class
FancyLogger.line([data, indent, level])Generic line logger
FancyLogger.row(\*args, \*\*kwargs)
FancyLogger.title(text[, level])Title, logged at info level
FancyLogger.section_header(text[, indent, level])Section header, logged at info level
FancyLogger.sub_header([text, indent, level])Sub header
FancyLogger.foot([text, time, item_count, ...])Footer, logged at info level
FancyLogger.data([field, value, unit, ...])Data line logger
FancyLogger.sep([level, length, indent])Horizontal separator, logged at info level
FancyLogger.table([cell_data, ...])Data table
FancyLogger.info([text, indent])Info line logger
FancyLogger.debug([text, indent])Debug line logger
FancyLogger.error([text, indent])Error line logger

9.2 FancyPrinter

dcase_util.processors.FancyPrinter

该类向控制台提供统一格式的状态打印。

用例:

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
ui = dcase_util.ui.FancyPrinter()
ui.title('title')
ui.section_header('section_header')
ui.sub_header('sub_header')
ui.foot('foot')
ui.line('line', indent=2)
ui.line('line', indent=4)
ui.line('line', indent=6)

# Data row with field and value
ui.data('data field', 'value', 'unit')

# Horizontal separator
ui.sep()

# Table
ui.table(cell_data=[[1, 2, 3], [1, 2, 3]])

# Faster way to create output tables without collecting data into one data structure.
ui.row('Header1', 'Header2', widths=[10,20], types=['float2','str20'])
ui.row('-','-')
ui.row(10.21231, 'String text')

输出:

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
title
section_header
========================================
=== sub_header ===
foot

line
line
line
data field : value unit
========================================
Col #0 Col #1
------ ------
1 1
2 2
3 3

Header1 | Header2 |
------- | ----------------- |
10.21 | String text |
类名功能
FancyPrinter([colors])Printer class
FancyPrinter.line([data, indent, level])Generic line logger
FancyPrinter.row(\*args, \*\*kwargs)
FancyPrinter.title(text[, level])Title, logged at info level
FancyPrinter.section_header(text[, indent, ...])Section header, logged at info level
FancyPrinter.sub_header([text, indent, level])Sub header
FancyPrinter.foot([text, time, item_count, ...])Footer, logged at info level
FancyPrinter.data([field, value, unit, ...])Data line logger
FancyPrinter.sep([level, length, indent])Horizontal separator, logged at info level
FancyPrinter.table([cell_data, ...])Data table
FancyPrinter.info([text, indent])Info line logger
FancyPrinter.debug([text, indent])Debug line logger
FancyPrinter.error([text, indent])Error line logger

9.3 FancyStringifier

dcase_util.processors.FancyStringifier

这个类可以用来产生统一格式的输出字符串。

类名功能
FancyStringifier()Fancy UI
FancyStringifier.title(text)Title
FancyStringifier.section_header(text[, indent])Section header
FancyStringifier.sub_header([text, indent])Sub header
FancyStringifier.foot([text, time, ...])Footer
FancyStringifier.line([field, indent])Line
FancyStringifier.formatted_value(value[, ...])Format value into string.
FancyStringifier.data([field, value, unit, ...])Data line
FancyStringifier.sep([length, indent])Horizontal separator
FancyStringifier.table([cell_data, ...])Data table
FancyStringifier.row(\*args, \*\*kwargs)Table row
FancyStringifier.class_name(class_name)Class name

10. Units(单元函数与类)

10.1 General function

dcase_util.utils. *

类名功能
get_class_inheritors(klass)得到从给定的类中继承的所有类
get_byte_string(num_bytes[, show_bytes])根据语言环境和IEC二进制前缀输出字节数
check_pkg_resources(package_requirement[, ...])
is_int(value)
is_float(value)
  • SuppressStdoutAndStderr
    dcase_util.utils.SuppressStdoutAndStderr
类名功能
SuppressStdoutAndStderr()Context manager to suppress STDOUT and STDERR
  • VectorRecipeParser
    dcase_util.utils.VectorRecipeParser
类名功能
VectorRecipeParser([delimiters, default_stream])

10.2 File

dcase_util.utils. *

类名功能
argument_file_exists(filename)参数文件检查器
filelist_exists(filelist)检查列表中的所有文件是否存在
posix_path(path)检查列表中的所有文件是否存在
  • Path
    dcase_util.utils.Path
类名功能
Path([path])路径的单元
Path.posix([path])将路径转换为POSIX格式
Path.posix_to_nt([path])将posix格式的路径转换为nt
Path.file_list([path, recursive, ...])获取文件列表
Path.exists([path])检查路径是否存在
Path.file_count([path])给定路径下的文件数量包括子目录
Path.size_bytes([path])给定路径下所有文件的总字节数
Path.size_string([path, show_bytes])给定路径下的所有文件的总数据大小以可读形式返回
Path.makedirs([path])创建给定的路径
Path.create(paths)创建给定的路径
  • ApplicationPaths
    dcase_util.utils.ApplicationPaths
类名功能
ApplicationPaths([parameter_container])应用程序路径的实用工具类,根据参数散列自动生成路径
ApplicationPaths.generate(path_base, structure)生成应用程序路径并将参数散列包含到路径中
ApplicationPaths.directory_name(prefix, ...)生成目录名称
ApplicationPaths.save_parameters_to_path(...)将参数保存到每个应用程序子目录
ApplicationPaths.construct_path(path_parts)基于路径部分生成所有组合
  • FileFormat
    dcase_util.utils.FileFormat
类名功能
FileFormat
FileFormat.detect(filename[, ...])检测文件格式
FileFormat.detect_based_on_filename(filename)基于文件名检测文件格式
FileFormat.detect_based_on_content(filename)使用python-magic检测基于内容的文件格式。
FileFormat.validate_label(label)根据此类已知的标签验证文件格式标签

10.3 Hash

类名功能
get_parameter_hash(params)获取给定参数字典的唯一哈希字符串(md5)
get_file_hash(filename)获取给定文件的唯一哈希字符串(md5)

10.4 Logging

类名功能
setup_logging修饰器类只允许执行一次
  • DisableLogger
    dcase_util.utils.DisableLogger
类名功能
DisableLogger

10.5 Math

dcase_util.utils.SimpleMathStringEvaluator

  • SimpleMathStringEvaluator
类名功能
SimpleMathStringEvaluator()简单的数学字符串评估器
SimpleMathStringEvaluator.eval(string)评估字符串中的数学

10.6 Timer

dcase_util.utils.Timer

类名功能
Timer()定时器类
Timer.start()启动计时器
Timer.stop()停止计时器
Timer.elapsed()返回自启动计时器以来的经过时间,以秒为单位
Timer.get_string([elapsed])以字符串格式获取已用时间

10.7 Validator

dcase_util.utils.FieldValidator

类名功能
FieldValidator
FieldValidator.process(field)测试字段
FieldValidator.is_empty(field)测试空字段
FieldValidator.is_number(field)测试数字字段
FieldValidator.is_audiofile(field)测试音频字段
FieldValidator.is_list(field)测试列表字段,有效分隔符[:;#]
FieldValidator.is_alpha(field[, length])测试长度为1的alpha字段

10.8 Example

dcase_util.utils.Example

一些示例数据,便于测试和辅导。

类名功能
Example()教程示例文件
Example.audio_filename()
Example.acoustic_scene_audio_filename()
Example.audio_container()
Example.event_metadata_container([filename])
Example.scene_metadata_container([filename])
Example.tag_metadata_container([filename])
Example.feature_container([filename])
Example.feature_repository([filename])
  • 0
    点赞
  • 2
    收藏
    觉得还不错? 一键收藏
  • 打赏
    打赏
  • 0
    评论

“相关推荐”对你有帮助么?

  • 非常没帮助
  • 没帮助
  • 一般
  • 有帮助
  • 非常有帮助
提交
评论
添加红包

请填写红包祝福语或标题

红包个数最小为10个

红包金额最低5元

当前余额3.43前往充值 >
需支付:10.00
成就一亿技术人!
领取后你会自动成为博主和红包主的粉丝 规则
hope_wisdom
发出的红包

打赏作者

flybirding10011

谢谢支持啊999

¥1 ¥2 ¥4 ¥6 ¥10 ¥20
扫码支付:¥1
获取中
扫码支付

您的余额不足,请更换扫码支付或充值

打赏作者

实付
使用余额支付
点击重新获取
扫码支付
钱包余额 0

抵扣说明:

1.余额是钱包充值的虚拟货币,按照1:1的比例进行支付金额的抵扣。
2.余额无法直接购买下载,可以购买VIP、付费专栏及课程。

余额充值