Data Collection

The data corpus for experiment comprises datasets of varied vocabulary sizes, which is created by collecting data during two main phases.

First, researchers invite three participants (two male, one female, the average age of 29.33 years) join the pilot study to investigate the feasibility of signal detection and to determine electrode positioning. The preliminary dataset recorded with the participants was binary, with the word labels being yes or no. The vocabulary set size gradually increases with more words. In sum, the data collected during the study has about 5 hours of internally vocalized text.

In the second phase, a data corpus was created to train a classifier (same 3 participants). The corpus has about 31 hours of internal speech text recorded in different sessions to be able to regularize the recognition model for session independence. The corpus comprises of multiple datasets. In one category, the word labels are numerical digits (0-9) along with fundamental mathematical operations (times, divide, add, subtract and percent) to facilitate externalizing arithmetic computations through the interface. They use the external trigger signal to slice the data into word instances. In each recording session, signals were recorded for randomly chosen words from a specific vocabulary set. This data is used to train the recognition model for various applications like World Clock, Calendar, Chess Game.

Midterm Update Summary

The progress with regard to the schedule

Technical challenges met, solved and remaining

I research on AlterEgo to find out how it to cognitively augment humans, change the

way people communicate with one another, and enable a discreet gateway to digital

information where the interaction is intrinsic rather than something extrinsic.

I have updated two main parts of my research

• A novel wearable architecture for a bidirectional silent speech device.

• How the neuromuscular input works.

will continually follow the schedule to figure out the third one.

• The utility of the device as a personal computing platform and interface.

Adjustment requested to the proposal

Luckily, there is no adjustment needed due to the proposal so far.

Anything else needs to be updated

More detail about the architecture, like the details in the convolutional neural network to classify into word labels. Possibly CNN knowledge needed for researching purpose.

Wearable Architecture: Aural Output & Wearable From Design

AURAL OUTPUT

AlterEgo system opens up a unique opportunity to achieve personalized two-way human-computer interaction takes place in a hidden and seamless way, where the elements of the interactions are in natural language. This potentially promotes complementary synergies between human users and machines, some of which can be outsourced to computers, while computing is still "intrinsic" for human users.

Once the internal utterance phrase is recognized, the computer translates the context based on the relevant application accessed by the user. Then the output computed by the application is then converted using a "text to speech" and transmitted to the user by hearing. Alterego uses bone conduction headphones as the aural output, so as to not interferes the user's sense of hearing.

WEARABLE FORM DESIGN

Mainly consider keeping the position of the electrodes in order to make the wearable system durable and can be used in regular settings.

1: The electrode does not move to maintain the signal consistency.

2: The electrode maintains the position between multiple target areas and the number of sessions the user wears and does not wear the device.

3: Although the electrodes must not move to stray force, the device purposed to adjust the position to different users' locations can be accessed from one device worn by multiple users.

To reach the goal, the device is designed as a wearable device and can be worn on the back of the head, the extension falls to the face. The signal from the target area is recorded and the band is 3D printed using photopolymer resin with a brass rod supporting it structurally, so as to maximize skin-electrode friction and minimize relative movement. The extensions are brass supports that provide rigidity to support electrodes while also being amenable to deliberate adjustment.

In addition, the extension is designed to be connected to electrodes in a modular frame that could be pulled out for further experimentation.