[Project] Measuring a listener's tolerance to the quality of audio files

[livesensus]

I am working with LiveSensus, a team comprised of 5 Computer Science and Computer Engineering students at the University of Illinois at Urbana-Champaign. LiveSensus is working on a project that measures a listener's tolerance to different audio issues that may come up during live streams, video conferencing call, and online lectures.

LiveSensus has created a machine learning program where users can rate audio clips based on their perception of the quality. At this point, the website is launched and we are seeking respondents interested in testing out our easy 5 minute demo. All data will remain confidential and anonymity will be ensured.

We've been really excited to get this project up and running in a Covid-19 world where many workers, students, etc., are transitioning to online alternatives. Please feel free to contact us with any questions and most of all any comments for improvement. Any insights would go a really long way!

You can access our website at: https://livesensus.com/

and the demo directly at: https://livesensus.com/survey.html

Has anyone tried or heard of anyone taking on a similar project?

[AlecSp]

Interestingly, I suspect the local accent used in many of the "clear" samples might render them rather less clear to a UK user...

[Eddy Deegan]

I'm familiar with talking to people in the US but I agree about the accent thing.

Also, it didn't work for me. After the practice snippets, the first 'real' question simply said "not ready" next to the play button, which was greyed out, and the page refused to do anything else so I gave up.

On on of the practice questions there is a typo by the way, where it says "high qualiy audio" instead of "high quality audio"

[resistorman]

This will be a highly sensitive and critical sample of the population, you know. And, as you can see, they barely tolerate ‘ mericans

[shufflebeat]

This will be a highly sensitive and critical sample of the population, you know. And, as you can see, they barely tolerate ‘ mericans

I'm not sure that's fair, most folks here wouldn't judge on a national or ethnic dimension, it actually wouldn't register with them unless addressed specifically.

Technical carelessness, on the other hand...

[RichardT]

This is a very worthwhile project, however the people who use this forum spend a lot of time listening to the quality of audio and thinking about issues of sound quality and I think there is a risk that they won’t respond to the sound quality of the clips in the same way as the general population.

[muzines]

I didn't have any problems with it, but I thought it was a bit weird in it's choices of examples.

I think what they are trying to evaluate is whether people find good quality audio but with missing or garbled content more or less annoying than lower quality audio where you can still make out all the words and follow the content.

[resistorman]

This will be a highly sensitive and critical sample of the population, you know. And, as you can see, they barely tolerate ‘ mericans

I'm not sure that's fair, most folks here wouldn't judge on a national or ethnic dimension, it actually wouldn't register with them unless addressed specifically.

Technical carelessness, on the other hand...

Just a bad joke

[AlecSp]

In the end, though, we don't know exactly what this survey is looking to assess.

Yes, the samples felt like an odd set. They could have been incredibly well designed, or a pile of pants. All we can do is contribute in good faith, trusting that it's competent.

Always interesting trying to second guess, mind...