Display and analysis

Display options

The model has a number of options for the output of the data for analysis. The two main options are an animation and an interactive plot. The animated display shows the width and location as an angle against loudness and frequency over time. The interactive plot shows either the width or location or a combination as an angle, together with loudness against time, for selected frequency bands and time segments.

Animated output

The animated output is designed to give an intuitive overview of the results for rapid and simple analysis. The location and width of each component is displayed as an angle in degrees on the x-axis. The results for each frequency band are plotted along the y-axis. The brightness of each part of the display indicates the loudness of the sound. This is all animated to show changes in the results over time. There is also the option to display a 'peak-hold' type meter that plots the extreme left and right positions. An example screenshot from this output is shown below, and there are links to examples of the full measurements.

[Screenshot of example animated output]

One of the advantages of this output type is that the original binaural signal can be combined with the video display to help the intuitive understanding of the result. However, the fact that it is animated makes it difficult to investigate the results in detail, especially rapid variations that may still be perceptually important. In addition, it is difficult to include these results in a written report. In view of this, a static interactive plot was developed.

Interactive plot

The static interactive plot contains two axes, both of which display results against time on the x-axis. The upper plot can display width, location, or a combination of these as an angle on the y-axis, as selected by the user. The user can also select which frequency bands to display, using the radio buttons on the left hand side of the user interface. The lower plot displays the loudness of the sounds, with the same frequency bands and time axis as the upper plot. The user can also select the time segment that is displayed.

Example measurement results are shown below: width, location and combination plots respectively.

Width

[Example width plot]

The example width plot above shows the variation in the perceived width over time for the frequency bands up to and including 1375 Hz. The result in each frequency band is shown individually as a different coloured line, with the same colour for a given frequency used for the two axes in the plot. If more detail is required, then a specific frequency band or time segment can be selected.

Location

[Example location plot]

The example location plot above shows the variation in the perceived location over time for the frequency bands up to and including 1375 Hz. Again, the result in each frequency band is shown individually as a different coloured line, with the same colour for a given frequency used for the two axes in the plot. If more detail is required, then a specific frequency band or time segment can be selected.

Combination

[Example combined width and location plot]

The example combination plot above shows the variation in the perceived location and width over time for the frequency bands up to and including 1375 Hz in a slightly different manner to the other plots. Each frequency band can be viewed individually as before, however, it is not practical to display the results for multiple frequency bands in this manner. Instead, a density plot is formed from the results, with the darker shades indicating a greater agreement between the frequency bands, and a lighter shade indicating a lesser agreement. In addition, the loudness of each frequency band affects the darkness of the plot, so that quieter and therefore perceptually less relevant results are suppressed compared to the louder and therefore perceptually more relevant results. This is why the plot becomes paler towards the right hand side as the sound decays.

Alternative output types

There are also a number of additional output options - it is intended that a large range of output options will be retained to allow detailed analysis of the measured results by the end user. A number of these are described below.

Full IACC plots over frequency, time and tau (τ)

In addition to the detail provided by the output over time and frequency, it is sometimes useful to view the full detail of the IACC calculation across the range of τ. This can be achieved by the use of an animation, as shown below. This shows the results of the measurement with τ on the x-axis and frequency on the y-axis, and the animation shows variations over time.

outputmov.wmv is a WindowsMedia movie showing an example of this type of output.

Plot of position and width

The final output option currently available is a plot of the predicted position (based on the interaural time difference) and the width as a radar plot. There are a number of options in how this can be created, including separating or combining results across time and frequency. This type of output will be more powerful once separate sources and reverberation can be identified and flagged as such, and once the results can be extended to the full 360 degree circle by including head movement in the binaural capture technique. Research is underway within the Institute of Sound Recording to enable this.

Single figure output

The simplest type of output is a single figure which can be used to quickly compare between a number of measured systems. As discussed previously, it is difficult to combine the results across frequency. It is also difficult to combine the results across time. Whilst we are aware of the usefulness of a single figure output, we believe that use of this will compromise the advantages to be gained from such a detailed analysis as is allowed by the other plot techniques shown above.

A simple output could be derived by taking either the maximum or mean values across time and frequency, though it is advised that such results are treated with caution. It is intended that further research into integration of the results across frequency and time will allow an automated single figure output. However, in the meantime, it is better for the user to qualitatively interpret the data to derive single figure results if necessary.