| Types of Error Reports |
|---|
EXPLANATION OF THIS SECTION
This section of the manual describes the various ways in which the user may view the results of jProductionCritic processing. Each of the error report types are described below.
OVERVIEW OF THE ERROR REPORTS
The purpose of using jProductionCritic is to detect technical production errors in audio files. The software can report such detected errors to the user in a variety of report formats, which are detailed below. The user may have anywhere from one to all of these reports generated for any given error analysis.
COMMAND LINE REPORT
This type of report is printed either directly to standard out (if jProductionCritic is being run via its command line arguments) or in the report window of the GUI (if jProductionCritic is being run via its GUI). The data is output together with jProductionCritic's processing status messages. The data presented in a Command Line Report is identical to that presented in a Text File Report.
The processing results for each audio file are presented separately and sequentially. The report for each audio file analyzed begins with a list of standard metadata (whether or not any production errors are detected for the given file), followed by the errors detected listed one by one. Errors of the same type are grouped together. For each error detected, the report includes the type of error, whether it is instantaneous or time spanning, its start time, its end time (if it is time spanning) and its severity.
TEXT FILE REPORT
This type of report is saved as .txt files (one per audio file analyzed). The data presented in a Command Line Report is identical to that presented in a Text File Report.
The report for each audio file analyzed consists of a single text file and begins with a list of standard metadata (whether or not any production errors are detected for the given file), followed by the errors detected listed one by one. Errors of the same type are grouped together. For each error detected, the report includes the type of error, whether it is instantaneous or time spanning, its start time, its end time (if it is time spanning) and its severity.
If a single audio file is analyzed via the single file option (e.g. using "-check" under the command line arguments), then a single text report will be generated at the path specified with the file name specified. If the contents of a directory are analyzed via the batch option (e.g. using "-batchcheck" under the command line arguments), then multiple report files will be generated at the path specified, one for each audio file analyzed. Each of the files will have an auto-generated name, consisting of the file name of the audio file from which the report was generated (with the extension stripped away), followed by an underscore, followed by the number of the report (starting at 1), followed by a ".txt" extension. A new directory is also created if one does not already exist at the specified directory path (although nested directories will not be created, and trying to do so will generate an error). Any files in the save directory with file names the same as those saved will automatically be overwritten.
HTML REPORT
This type of report is saved as formatted HTML files (one per audio file analyzed).
The report for each audio file analyzed consists of a single HTML file and begins with a list of standard metadata (whether or not any production errors are detected for the given file), followed by the errors detected listed one by one. Errors of the same type are grouped together. For each error detected, the report includes the type of error, whether it is instantaneous or time spanning, its start time, its end time (if it is time spanning) and its severity.
If a single audio file is analyzed via the single file option (e.g. using "-check" under the command line arguments), then a single HTML report will be generated at the path specified with the file name specified. If the contents of a directory are analyzed via the batch option (e.g. using "-batchcheck" under the command line arguments), then multiple report files will be generated at the path specified, one for each audio file analyzed. Each of the files will have an auto-generated name, consisting of the file name of the audio file from which the report was generated (with the extension stripped away), followed by an underscore, followed by the number of the report (starting at 1), followed by a ".html" extension. A new directory is also created if one does not already exist at the specified directory path (although nested directories will not be created, and trying to do so will generate an error). Any files in the save directory with file names the same as those saved will automatically be overwritten.
AUDACITY LABEL TRACKS
This type of report is saved as Audacity Label Track files (one per audio file analyzed). These are metadata tracks that the popular and free Audacity DAW software displays alongside audio and MIDI tracks. They can be very useful for visually demonstrating to users where errors occur in a waveform or spectrogram. Audacity was chosen in particular because it is free and thus accessible to all users, whether or not they used it to prepare the audio being checked for errors.
This Audacity label file includes a label for each error detected. Each line of this file (which is essentially a text file) contains a single line for each label, and each such line consists of the start time of the error (in seconds), followed by a tab, followed by the end time of the error (in seconds), followed by a tab, followed by a label indicating the error type and its severity.
If a single audio file is analyzed via the single file option (e.g. using "-check" under the command line arguments), then a single Audacity label track file will be generated at the path specified with the file name specified. If the contents of a directory are analyzed via the batch option (e.g. using "-batchcheck" under the command line arguments), then multiple report files will be generated at the path specified, one for each audio file analyzed. Each of the files will have an auto-generated name, consisting of the file name of the audio file from which the report was generated (with the extension stripped away), followed by an underscore, followed by the number of the report (starting at 1), followed by a ".txt" extension. A new directory is also created if one does not already exist at the specified directory path (although nested directories will not be created, and trying to do so will generate an error). Any files in the save directory with file names the same as those saved will automatically be overwritten. Note that automatically generated Audacity label track file names are given the .txt extension, which could cause them to overwrite .txt report files set to be generated in the same directory, so it is wise to save text file reports and Audacity label track reports in separate directories.
ACE XML REPORT
This type of report is saved as a single ACE XML 1.1 Classifications File. This is the native format used by the jMIR music information retrieval research software, of which jProductionCritic is a part. Although not directly applicable to the educational context targeted by jProductionCritic (since ACE XML is intended primarily for machine learning applications), these ACE XML reports could be helpful to MIR researchers who might want to use the output of jProductionCritic in research involving machine learning. These formats can also be useful when performing experimental validations.
The ACE XML file produced details the detected errors for all analyzed audio files. Each audio file is treated as a separate instance, and each detected error is treated as a time-stamped classification for its instance. Each such classification includes a class label identifying the error type as well as a class label indicating its severity. Each instance is also labeled with all available metadata.
If a file already exists at the save path specified (e.g. using "-reportacexml" under the command line arguments), then it will automatically be overwritten. Note that, unlike the other report types, only one file is generated, regardless of whether the check or batch check extraction option is used. Also, the path specified must always refer to a path, not a directory, and no directory in the path will be created if it does not already exist.
WEKA ARFF REPORT
This type of report is saved as Weka ARFF files (one per audio file analyzed). This is the file format used by the popular and free Weka machine learning software. Although not directly applicable to the educational context targeted by jProductionCritic, these ARFF reports could be helpful to MIR researchers who might want to use the output of jProductionCritic in research involving machine learning. These formats can also be useful when performing experimental validations.
Each Weka ARFF file details the detected errors for a single analyzed audio file. The attributes for each instance are the start time (in seconds), end time (in seconds), severity and (as the class attribute) the error identifier (with whitespace removed). The candidate classes are all of the error types checked for (with whitespace removed). Note that this is an unusual type of data assignment to use in a Weka ARFF file, but is the best way to preserve the information output by jProductionCritic given the limitations of the ARFF format.
If a single audio file is analyzed via the single file option (e.g. using "-check" under the command line arguments), then a single ARFF file will be generated at the path specified with the file name specified. If the contents of a directory are analyzed via the batch option (e.g. using "-batchcheck" under the command line arguments), then multiple report files will be generated at the path specified, one for each audio file analyzed. Each of the files will have an auto-generated name, consisting of the file name of the audio file from which the report was generated (with the extension stripped away), followed by an underscore, followed by the number of the report (starting at 1), followed by a ".arff" extension. A new directory is also created if one does not already exist at the specified directory path (although nested directories will not be created, and trying to do so will generate an error). Any files in the save directory with file names the same as those saved will automatically be overwritten.