Prerequisites for using EmotivPipeline are that the user must: a) be a current EmotivPRO license holder and b) be syncing data with Emotiv’s cloud servers. In order to see the EmotivPipeline features, you need to enable them within your EmotivPRO Settings. Go to the EmotivPRO settings (click the GEAR icon on the right-hand side of the top menu bar) and select the checkbox next to “Activate EmotivPipeline”.
In the recordings tab of EmotivPRO, select all the recordings that you would like to process in the cloud. Once you have activated EmotivPipeline and select one or more recordings, the “Process Data” button will appear in the top right hand corner of the screen next to the ”Export Data” and “Delete Locally” options.
Clicking on the “Process Data” button will redirect you to the EmotivPipeline web page. You will need to login using your EmotivID.
Types of Processing
Currently there are two types of processing: Basic processing, which consists of re-referencing and high-pass filtering, and Custom Band Power calculation, where you can specify the frequency bands that you want to calculate. Below we describe each setting in more detail.
Filtering and Re-referencing Parameters:
Slewrate limit: [default 30 microvolts] Limits the change in the EEG signal between successive samples to a maximum value. Very large step changes from one sample to the next sometimes occur due to electrostatic discharge of the subject, mechanical impacts on the headset or other artefacts. The effects of these voltage steps on filtered signals and FFT epochs can artificially persist for a long time after the event. Limiting the extent of a step change reduces its impact without significantly affecting the actual EEG signal components. Re-referencing: The signal can be re-referenced to an aggregate of all the signals at each sample. Options are [“mean”, “median”, “interquartile mean”, “None”]
Filter Coefficients: Enter the filter coefficients of a high pass filter to be performed where the coefficients a and b satisfy:
a*yn + a*yn-1 +…+ a[l]*yn-l = b*xn + b*xn-1 +…+ b[l]*xn-l
Where xn is the nth input data point and yn is the nth output of the filter. The default filter is a second order butterworth filter with a 3dB point at 0.5Hz at 128Hz.
Fourier transform window size: default: 256
Sliding Window step size: default: 64
Band frequencies: Enter the number of bands desired with the frequencies desired. The frequencies include the lower value and exclude the upper value. For example the bands 4 to 8 Hz and 8 to 12 Hz do not overlap and the Fourier component for 8Hz is only included in the second band.