Closed jackdolgin closed 4 years ago
Hi @jackdolgin, thanks for reaching out, and we're looking forward to your pull request. @paxtonfitzpatrick, @KirstensGitHub, or @andrewheusser might have more insight into your question, but my memory is that the microphone only appears for a few seconds to prompt you to start speaking. During the blank screen it should still be recording. Is that the behavior you're experiencing, or is the recording stopping early too? The other thought I have is that you may have accidentally set the debug flag to true somewhere, since that will shorten the recording time too.
Hi @jeremymanning, thanks for the quick reply! I think I just figured out a mistake I made that affects the blank screen part—I was running the debug link (http://0.0.0.0:22362/ad?etc) rather than localhost:22362. Now the experiment ends on its own and appears to record the data. I'm still receiving the same error message in server.log
, though, and the experiment is still only two words long. Are only two words ('saw' and 'chisel') supposed to be remembered, or are there supposed to be more words (perhaps from speech-context.csv
or wordpool.csv
)? Thanks again!
Re: the list length and number of lists, both of those are adjustable (see config.js). You can adjust the list length, number of lists, recording duration, and other experimental parameters by tweaking the settings in that file.
The wordlist.csv file provides the full word pool (from which presented words are drawn), and the speech-context.csv file contains the "prior" that the speech-to-text engine uses (i.e. when there are ambiguities in the speech, words from the speech context file are preferred over similarly likely words not listed in that file). You can change one or both of those files to suit your experiment's needs. (But the speech-context.csv file should contain all of the words listed in the wordlist.csv file.)
It looks like this issue is now resolved, but please re-open it (or open a new issue) if you're still having trouble! (And thanks again for your pull request!)
Thank you, works perfectly now! Looks like that was a mistake on my end.
Hi,
Great package and super useful! I'm going to submit a pull request (I'm sort of new to GitHub but hopefully that's the right thing to do) with a couple updates to the repo to account for new semantics in using Google's API in the
custom.py
file, but even accounting for that change, I'm running into the following error running the experiment. As soon as I click that the participant has "Yes, they have signed" the consent form, the terminal displays "[psiTurk server:on mode:sdbx #HITs:0]$ creating audio folder... " and theserver.log
displays the error below. I'm able to run a sample of the experiment—just the appearance of the first two words, followed by instructions to recall all the words I remember (which are only two since I only saw two) and then after the red microphone appears for ~5 seconds the screen goes blank. Neither the terminal norserver.log
have any additional messages. Any guidance would be much appreciated!