After completing the RC experiment and the resynthesis adjustment, we present subjects with a set of subjective judgement questions. In a randomized order, we play them 1) their reconstruction with no adjustments rescalled to between -20 and 0 dB, 2) their reconstruction after applying the adjustment parameters they chose in the adjust resynthesis section, and 3) white noise. For each sound, we ask them to rate on a scale of 1-7 their answer to the statement: "this sounds like my tinnitus" OR "this sounds like the target signal". You can see the screens here in the ones labelled "compare". The code to run the questions is here. Note that it has some extra pieces (the survey questions, the follow up version differences) that we won't have to worry about.
IMPLEMENTATION:
Similar to the adjustment routine, these questions should appear right after the adjustment piece has finished.
These questions should also be available from the main page in case the window is closed.
Usually these questions are repeated at least once. So, there should be some counter to determine how many times it has been run and if it's hit the set amount. I'm not sure where that should be set. I don't think it will change by experiment. Maybe that's some server-side parameter that's just baked into the code. Or maybe it should be allowed infinite repeats. We can discuss.
The responses to each stimulus should be saved and clearly labeled. That is, not just i.e., Q1: 4, Q2: 2, Q3: 6, but something like standard_resynth: 4, white_noise: 2, adjusted_resynth: 6. And of course associated with a user, experiment, instance, etc.
It might be nice if there was the option to add more questions, but that's not really necessary.
OVERVIEW:
After completing the RC experiment and the resynthesis adjustment, we present subjects with a set of subjective judgement questions. In a randomized order, we play them 1) their reconstruction with no adjustments rescalled to between -20 and 0 dB, 2) their reconstruction after applying the adjustment parameters they chose in the adjust resynthesis section, and 3) white noise. For each sound, we ask them to rate on a scale of 1-7 their answer to the statement: "this sounds like my tinnitus" OR "this sounds like the target signal". You can see the screens here in the ones labelled "compare". The code to run the questions is here. Note that it has some extra pieces (the survey questions, the follow up version differences) that we won't have to worry about.
IMPLEMENTATION:
Q1: 4, Q2: 2, Q3: 6
, but something likestandard_resynth: 4, white_noise: 2, adjusted_resynth: 6
. And of course associated with a user, experiment, instance, etc.