comparison docs/SMC15/smc2015template.tex @ 79:a2fd892ab6b9

Paper: expanded introduction and list of options

 others from the list show real-time captured audio processing such as room reverberation tools and a phase vocoder from the system microphone. The BBC Radiophonic Workshop shows effects used on famous TV shows such as Doctor Who, being simulated inside the browser \cite{bbcradiophonics}. %http://webaudio.prototyping.bbc.co.uk/
 Another example is the BBC R\&D automatic compressor which applies a dynamic range compressor on a radio station which dynamically adjusts the compressor settings to match the listener's environment \cite{mason2015compression}.\\ % The paper for this has not been released yet by AES...
 
 This work is based in part on the APE audio perceptual evaluation interface for MATLAB \cite{deman2014b}. An important drawback of this toolbox is the need to have MATLAB to create a test and even to run (barring the use of an executable generated by MATLAB), and limited compatibility with both earlier and newer versions of MATLAB, which makes it hard to maintain. On the other hand, a web application generally has the advantage of running in most browsers on most applications, and we present a tool that requires no specialised software or even programming knowledge to set up. 
 
-BRECHT: We present a browser-based perceptual evaluation tool for audio that ... \\
-what can we do with this: multiple stimulus perceptual evaluation \cite{bech}, commenting, rating, ranking, (no AB) - can also be single \\ % also in abstract
-goal, what are we trying to do? \\
+[TO ADD: other interfaces for perceptual evaluation of audio, browser-based or not!] % IMPORTANT
+
+We present a browser-based perceptual evaluation tool from which any kind of multiple stimulus audio evaluation tool where subjects need to rank, rate, select, or comment on different audio samples can be built. 
+We also include an example of the multiple stimulus user interface included with the APE tool \cite{deman2014b}, which presents the subject with a number of axes on which a number of markers, corresponding with audio samples, can be moved to reflect any subjective quality, as well as corresponding comment boxes. 
+However, other graphical user interfaces can be put on top of the engine that we provide with minimal or no modifications. Examples of this are the MUSHRA test \cite{mushra}, single or multiple stimulus evaluation with a two-dimensional interface (such as valence and arousal dimensions), or simple annotation (using free-form text, check boxes, radio buttons or drop-down menus) of one or more audio samples at a time. 
+In case of for instance method of adjustment, where the audio is processed by the user \cite{bech}, or of AB test, where the interface does not show all audio samples to be evaluated at once, the back end of the tool needs to be modified as well. 
+
 
 DAVE: justify:
 \begin{itemize}
 \item Easy deployment - no propitiatory software required
 \item Entirely scalable
 \item Trial multiple participants simultaneously
 \item Intuitive environment - everyone uses the internet
 \item Can be hosted and run from single location with participants worldwide
 \end{itemize}
 
-
-
-
-see \cite{deman2014b}: requirements informed by research on music production (see my work and that of others' in the group), such as randomisation, playback of high quality audio, some degree of flexibility in terms of configuration, ... \\
+[TO ADD: in the following sections, we describe the interface in more detail, discuss the implementation, and go over design considerations and flexibility ... ]
 
 
 \section{Interface}\label{sec:interface}
 
-At this point, we have implemented the interface of the MATLAB-based APE Perceptual Evaluation for Audio toolbox \cite{deman2014b}, which shows one marker for each simultaneously evaluated audio fragment on one or more horizontal axes (to rate/rank the respective fragments), as well as a comment box for every marker, and any extra text boxes for extra comments. See \ref{fig:interface} for an example of the interface, with eleven fragments and one axis. However, the back end of this test environment allows for many more established and novel interfaces for listening tests, particularly ones where the subject only assesses audio without manipulating it (i.e. method of adjustment, which would require additional features to be implemented). 
+At this point, we have implemented the interface of the MATLAB-based APE Perceptual Evaluation for Audio toolbox \cite{deman2014b}, which shows one marker for each simultaneously evaluated audio fragment on one or more horizontal axes, that can be moved to rate or rank the respective fragments in terms of any subjective quality, as well as a comment box for every marker, and any extra text boxes for extra comments. See \ref{fig:interface} for an example of the interface, with eleven fragments and one axis. %? change if a new interface is shown
+
+However, the back end of this test environment allows for many more established and novel interfaces for listening tests, particularly ones where the subject only assesses audio without manipulating it (i.e. method of adjustment, which would require additional features to be implemented). 
 
 \begin{figure*}[htbp]
 \begin{center}
 \includegraphics[width=0.9\textwidth]{interface.png}
 \caption{Example of interface, with 1 axis and 11 fragments}

 
 The HTML file loads the \texttt{core.js} file along with a few other ancillary files (such as the jQuery JavaScript extensions)% should we cite jQuery.... https://jquery.com/
 , at which point the browser JavaScript begins to execute the on-page instructions, which gives the URL of the test setup XML document (outlined in Section \ref{sec:setupresultsformats}). \texttt{core.js} parses this document and executes the function in \texttt{ape.js} to build the web page with the given audio files. The reason for separating these two files is to allow for further interface designs (such as MUSHRA \cite{mushra} or A-B tests \cite{bech}) to be used, which would still require the same underlying core functions outlined in \texttt{core.js}, see also Section \ref{sec:interface}.
 
 The \texttt{ape.js} file has several main functions but the most important are \textit{loadInterface(xmlDoc)}, \textit{loadTest(id)}, \textit{pageXMLSave(testId)} and \textit{interfaceXMLSave()}. \textit{loadInterface(xmlDoc)} is called to decode the supplied project document in respect for the interface specified and define any global structures (such as the slider interface). It also identifies the number of pages in the test and randomises the order, if specified to do so. This is the only madatory function in any of the interface files as this is called by \texttt{core.js} when the document is ready. The design style is such that \texttt{core.js} cannot 'see' any interface specific functions and therefore cannot assume any are available. Therefore the \textit{loadInterface(xmlDoc)} is very important to setup the entire test environment. It can be assumed that the interface files can `see' the \texttt{core.js} file and can therefore not only interact with it, but also modify it.
+
 Each test page is loaded using \textit{loadTest(id)} which performs two major tasks: to populate the interface with the slider elements and comment boxes; and secondly to instruct the \textit{audioEngine} to load the audio fragments and construct the backend audio graph. The markers on the slider at the top of the page are positioned randomly, to minimise the bias that may be introduced when the initial positions are near the beginning, end or middle of the slider. While another approach is to place the markers outside of the slider bar at first and have the subject drag them in, the authors believe this doesn't encourage careful consideration and comparison of the different fragments as the implicit goal of the test becomes to audition and drag each fragment in just once, rather than to compare all fragments rigorously.
+
+\textit{loadTest(id)} in \texttt{ape.js} also instructs the audio engine in \texttt{core.js} to create the \textit{audioObject} These are custom audio nodes, one representing each audio element specified in each page.
 They consist of a \textit{bufferSourceNode} (a node which holds a buffer of audio samples for playback) and a \textit{gainNode}. There are various functions applied depending on metric collection which record the interaction with the audio element. These nodes are then connected to the \textit{audioEngine} (itself a custom web audio node) containing a \textit{gainNode} (where the various \textit{audioObject} connect to) for summation before passing the output to the \textit{destinationNode}, a permanent fixed node of the Web Audio API created as the master output where the browser then passes the audio information to the system sound device. % Does this now make sense?
 % audio object/audioObject/Audio Object: -- should always be audioObject if talking about the JavaScript object, otherwise should say audio element or audio fragment.
+
 When an \textit{audioObject} is created, it is given the URL of the audio sample to load. This is downloaded into the browser asynchronously using the \textit{XMLHttpRequest} object. This downloads any file into the JavaScript environment for further processing which is particularly useful for the Web Audio API because it supports downloading of files in their binary form for decoding. Once downloaded the file is decoded using the Web Audio API offline decoder. This uses the browser available decoding schemes to decode the audio files into raw float32 arrays, which are in turn passed to the relevant \textit{audioObject} for playback.
 
 Once each page of the test is completed, identified by pressing the Submit button, the \textit{pageXMLSave(testId)} is called to store all of the collected data until all pages of the test are completed. After the final test and any post-test questions are completed, the \textit{interfaceXMLSave()} function is called. This function generates the final XML file for submission as outlined in Section \ref{sec:setupresultsformats}.
 
 Browsers support various audio file formats and are not consistent in any format. Currently the Web Audio API is best supported in Chrome, Firefox, Opera and Safari. All of these support the use of the uncompressed WAV format. Although not a compact, web friendly format, most transport systems are of a high enough bandwidth this should not be a problem. Ogg Vorbis is another well supported format across the 4 supported major desktop browsers, as well as MP3 (although Firefox may not support all MP3 types) \cite{mozdevSupportedMedia}. %https://developer.mozilla.org/en-US/docs/Web/HTML/Supported_media_formats
 One issue of the Web Audio API is that the sample rate is assigned by the system sound device, rather than requested and does not have the ability to request a different one. % Does this make sense? The problem is across all audio files.
  Therefore, the default operation when an audio file is loaded with a different sample rate to that of the system is to convert the sample rate. To provide a check for this, the desired sample rate can be supplied with the setup XML and checked against. If the sample rates do not match, a browser alert window is shown asking for the sample rate to be correctly adjusted. 
+As this happens before any loading or decoding of audio files, the system will only fetch files as soon as the system's sample rate meets any requirements, avoiding requests for large files until they are actually needed.
+
+%During playback, the playback nodes loop indefinitely until playback is stopped. The gain nodes in the \textit{audioObject}s enable dynamic muting of nodes. When a bar in the sliding ranking is clicked, the audio engine mutes all \textit{audioObject}s and un-mutes the clicked one. Therefore, if the audio samples are perfectly aligned up and of the same sample length, they will remain perfectly aligned with each other.
+% Don't think this is relevant anymore
+
+\section{Setup and results formats}\label{sec:setupresultsformats}
+
+[DAVE: somewhere add input file]
+
+Setup and the results both use the common XML document format to outline the various parameters. The setup file determines which interface to use, the location of audio files, how many pages and other general setup rules to define the testing envrionment. Having one document to modify allows for quick manipulation in a `human readable' form to create new tests, or adjust current ones, without needing to edit multiple web files. % I mean the .js and .html files, though not sure if any better.
+The setup document has several defined nodes and structure which are documented with the source code. For example there is a section for general setup options where the pre-test and post-test questions and statements are defined:
+
+\texttt{<question id="location" mandatory="true"> Please enter your listening location \\ </question>}
+
+From the above example it can be seen that a question box should be generated, with the id 'location' and it is mandatory to answer. The question is in the PreTest node meaning it will appear before any testing will begin. When the result for the  entire test is shown, then this will appear in the PreTest node of the response with the id 'location' allowing it to be found easily. This outlines the importance of having clear and meaningful ID values. Pre- and post-test dialog boxes allow for comments or questions to be presented before or after the test, to convey listening test instructions, gather information about the subject, listening environment, and overall experience of the test.
+ 
+[BRECHT to say we try to cater to as many audiences as possible, simple yet tweakable]
+
+Further options in the setup file are: 
+
+\begin{itemize}
+\item \textbf{Snap to corresponding position}: When this is enabled, and a fragment is playing, the playhead skips to the same position in the next fragment that is clicked. If it is not enabled, every fragment is played from the start. 
+\item \textbf{Loop fragments}: Repeat current fragment when end is reached, until the `Stop audio' or `Submit' button is clicked. 
+\item \textbf{Comments}: Displays a separate comment box for each fragment in the page.
+\item \textbf{General comment}: One comment box, additional to the individual comment boxes, to comment on the test or a feature that some or all of the fragments share. 
+\item \textbf{Resampling}: When this is enabled, tracks are resampled to match the subject's system's sample rate (a default feature of the Web Audio API). When it is not, an error is shown when the system does not match the requested sample rate. 
+\item \textbf{Randomise page order}: Randomises the order in which different `pages' are presented. % are we calling this 'pages'?
+\item \textbf{Randomise fragment order}: Randomises the order and numbering of the markers and comment boxes corresponding with the fragments. This permutation is stored as well, to be able to interpret references to the numbers in the comments (such as `this is much [brighter] then 4'). 
+\item \textbf{Require playback}: Require that each fragment has been played at least once, if not in full. 
+\item \textbf{Require full playback}: If `Require playback' is active, require that each fragment has been played in full. 
+\item \textbf{Require moving}: Require that each marker is moved (dragged) at least once. 
+\item \textbf{Require comments}: This option allows requiring the subject to require a comment for each track. 
 \item \textbf{Repeat test}: Number of times test should be repeated (none by default), to allow familiarisation with the content and experiment, and to investigate consistency of user and variability due to familiarity. In the setup, each 'page' can be given a repeat count. These are all gathered before shuffling the order so repeated tests are not back-to-back if possible.
-% explanation on how this is implemented? -- Is this better?, Nick
+\item \textbf{Returning to previous pages}: Indicates whether it is possible to go back to a previous `page' in the test. 
+\item \textbf{Lowest rating below [value]}: To enforce a certain use of the rating scale, it can be required to rate at least one sample below a specified value. 
+\item \textbf{Highest rating above [value]}: To enforce a certain use of the rating scale, it can be required to rate at least one sample above a specified value. 
+\item \textbf{Reference}: Allows for a separate sample (outside of the axis) to be the `reference', which the subject can play back during the test to help with the task at hand \cite{mushra}. 
+\item \textbf{Hidden reference}: Whether or not an explicit `reference' is provided, the `hidden reference' should be rated above a certain value \cite{mushra} - this can be enforced. 
+\item \textbf{Hidden anchor}: The `hidden anchor' should be rated lower than a certain value \cite{mushra} - this can be enforced. 
+
+\end{itemize}
+
+When one of these options is not included in the setup file, they assume a default value. 
+
+% loop, snap to corresponding position, comments, 'general' comment, require same sampling rate, different types of randomisation
+
+The results file is dynamically generated by the interface upon clicking the `Submit' button. This also executes checks, depending on the setup file, to ensure that all tracks have been played back, rated and commented on. The XML output returned contains a node per audioObject and contains both the corresponding marker's position and any comments written in the associated comment box. The rating returned is normalised to be a value between 0 and 1, normalising the pixel representation of different browser windows.
+
+The results will also contain information collected by any defined pre/post questions. These are referenced against the setup XML by using the same ID so readable responses can be obtained. Taking from the earlier example of setting up a pre-test question, an example reponse would be shown as the following.
+
+\texttt{<comment id="location"> Queen Mary \\University of London </comment>}
+
+Each page of testing is returned with the results of the entire page included in the structure. One `audioElement' node is created per audio fragment per page, along with its ID. This includes several child nodes including the value holding the rating between 0 and 1, and any metrics collected. These include how long the element was listened for, the initial position, boolean flags if the element was listened to, if the element was moved and if the element comment box had any comment. Furthermore, each user action (manipulation of any interface element, such as playback or moving a marker) is logged along with a the corresponding time code and stored or sent along with the results.
+
+Future development will also evolve to include any session data, such as the browser the tool was used in. Currently the results files are downloaded on the user side of the browser as a .xml file to be manually returned. However the end goal is to allow the XML files to be submitted over the web to a receiving server to store them, allowing for automated collection.
+
+ % right? 
+
+%Here is an example of the setup XML and the results XML: % perhaps best to refer to each XML after each section (setup <> results)
+% Should we include an Example of the input and output XML structure?? --> Sure. 
+
+An example of the returned \textit{audioElement} node in the results XML file is as follows.
+
+\texttt{<audioelement id="8"> \\
+<comment> \\
+<question>Comment on track 0</question> \\
+<response> The drums were punchy </response> \\
+</comment> \\
+<value> 0.25169491525423726 </value> \\
+<metric> \\
+<metricresult id="elementTimer"> \\ 2.3278004535147385< /metricresult> \\
+<metricresult id="elementTrackerFull"> \\
+<timepos id="0"> \\
+<time>1.7937414965986385</time> \\
+<position>0.41694915254237286</position> \\
+</timepos> \\
+<timepos id="1"> \\
+<time>2.6993197278911563</time> \\
+<position>0.45847457627118643</position> \\
+</timepos> \\</metricresult> \\
+<metricresult id="elementInitialPosition"> 0.47796610169491527 </metricresult> \\
+<metricresult id="elementFlagListenedTo"> true< /metricresult> \\
+<metricresult id="elementFlagMoved"> true </metricresult> \\
+</metric> \\
+</audioelement>}
+
+As can be seen, the parent tag \texttt{audioelement} holds the id of the element passed in from the setup document. The first child element is \texttt{comment} and holds both the question shown and the response from the comment box inside.
+The child element \texttt{value} holds the normalised ranking value. Next comes the metric node structure, there is one \texttt{metricresult} node per metric event collected. The id of the node identifies the type of data it contains. For example, the first holds the id \textit{elementTimer} and the data contained represents how long, in seconds, the audio element was listened to. The next holds the id \textit{elementTrackerFull} and contains a pair of elements per entry. This represents the entire movement of the elements' slider giving the time the event took place in seconds from when the current test page started, and the new position. In our example there are three \texttt{timepos} children with their id representing their order. There is one of these \texttt{audioelement} tags per audio element outlined on each test page. 
+
+\section{Conclusions and future work}\label{sec:conclusions}
+
+In this paper we have presented an approach to creating a browser-based listening test environment that can be used for a variety of types of perceptual evaluation of audio. 
+Specifically, we discussed the use of the toolbox in the context of assessment of preference for different production practices, with identical source material. 
+The purpose of this paper is to outline the design of this tool, to describe our implementation using basic HTML5 functionality, and to discuss design challenges and limitations of our approach. % or something
+
+% future work
+Further work may include the development of other common test designs, such as MUSHRA \cite{mushra}, AB, ABX and method of adjustment tests. 
+In addition, [...]. 
+[BRECHT to correct and expand]
+
+from community
+
+%\begin{itemize}
+%\item Options for MUSHRA style experiment with vertical slide per track
+%\item APE style experiment where all tracks are on a single horizontal axis % isn't that what we're doing now? 
+%\item AB test
+%\item ABX test
+%\item Method of adjustment tests
+%\end{itemize}
+
+
+The source code of this tool can be found on \url{code.soundsoftware.ac.uk/projects/webaudioevaluationtool}. 
+
+
+%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
+%bibliography here
+\bibliography{smc2015template}
+
+
+
+\end{document}
+
+
+
+% RUBBISH
+
+%\subsection{Equations}
+%Equations of importance, 
+%or to which you refer later,
+%should be placed on separated lines and numbered.
+%The number should be on the right side, in parentheses.
+%\begin{equation}
+%E=mc^{2+\delta}.
+%\label{eq:Emc2}
+%\end{equation}
+%Refer to equations like so:
+%As (\ref{eq:Emc2}) shows, 
+%I do not completely trust Special Relativity.
+%
+%\subsection{Figures, Tables and Captions}
+%\begin{table}[t]
+% \begin{center}
+% \begin{tabular}{|l|l|}
+%  \hline
+%  String value & Numeric value \\
+%  \hline
+%  Hej SMC  & 2015 \\
+%  \hline
+% \end{tabular}
+%\end{center}
+% \caption{Table captions should be placed below the table, exactly like this,
+% but using words different from these.}
+% \label{tab:example}
+%\end{table}
+
+%\begin{figure}[t]
+%\figbox{
+%\subfloat[][]{\includegraphics[width=60mm]{figure}\label{fig:subfigex_a}}\\
+%\subfloat[][]{\includegraphics[width=80mm]{figure}\label{fig:subfigex_b}}
+%}
+%\caption{Here's an example using the subfig package.\label{fig:subfigex} }
+%\end{figure}
+
+
+
+
+
+%\begin{acknowledgments}
+%You may acknowledge people, projects, 
+%funding agencies, etc. 
+%which can be included after the second-level heading
+%``Acknowledgments'' (with no numbering).
+%\end{acknowledgments} 
+