Apologies if my thoughts aren’t a bit more structured (and are a bit hodge podge), but I know if I wait till they are I’ll probably never get them sent.
I was using the site and had various thoughts…
Accent seems very important for this entire use case.
Are we gathering location data (IP) to assist with non-filled profile data?
For “listening” section, are we preferring to play similar accents to those who marked their accent in their profile?
I was playing clips and being from the U.S. could readily mark U.S. accents, as well as British accents, as either right or wrong. However some other accents were more difficult to understand. I listened to one like this with an Indian accent that other Indians may find perfectly reasonable. My concern is that some speakers are more likely to mark other speakers as wrong/bad.
This would have the effect of missing out on good audio as well as allowing the majority accent users on the site to set the bar for what’s acceptable for all accents.
This point was driven home for me recently when I heard a voice assistant on an Indian-made phone. The intonations alone made me have to work to understand it.
On that front, let’s make filling out the profile a part of “the path” that users take. Perhaps after the first 3 listens or speaks, we prompt users (even optionally, don’t need to stop them) to fill out the profile. Say reasons why it’s helpful (see the paragraph below).
Looking at the page for the profile, I ask myself, “Why should I fill this out?” We should make this section compelling; give it a reason. Something to the effect: “Telling us some basic data about yourself helps us to better…”
Listening: Let’s consider placing a checkbox to the right of the play button to allow for autoplay. So, if checked, as soon as I vote, the next clip will load and play. I personally found it annoying to have to click play again and again when I was “binging”.
Assure listeners that a thump or click at the beginning or end of recordings is ok. This doesn’t need to be a permanent fixture. Perhaps only displays during the first listen phase.
For the Listen section: “Yes!” and “Nope” should be disabled until the audio has played at least once. I have since discovered that an issue was already filed on this, but would add my +1.
Also, after the first play, they should remain enabled for a second playthrough.
Are we using Opus? (if not, why?)
What about audio levels? I’ve heard some recordings that were too faint for me to confidently say “yes” (I could barely hear them). Should we attempt to compensate for these and turn up the level? Some may become too distorted, but perhaps that’s ok as they’ll get marked “nope”.
Perhaps we should add something to the recording interface that analyzes and notifies users if the overall recording is too quiet.
Let’s keep track of our recording needs vs listening needs and recommend to users just arriving based on our current need. (If most people just want to record, for instance, that’s going to leave us very imbalanced.)
If we’re doing some amount of IP geo location, we can take this into account too - play South African accents for (likely) South African visitors - even if they haven’t filled out their profile.
About the Profile page… Why is the email above the demographic data? Why are you asking for my email?
I can make guesses, but I don’t think people are going to feel very compelled to share as it is. Also, as a non-requirement, I think we should consider putting the more important demographic data section above it.