Ableton Forum

Hello Everyone!

I'm a researcher currently studying for a PhD in the field of Psychoacoustics within the Applied Psychoacoustics Laboratory at the University of Huddersfield, UK.

As part of my research I’m looking into the perception of ‘punch’. In previous research we have developed a model that is able to measure the perception of punch, similarly to how a loudness meter is able to measure the perception of loudness, with promising results.

We’re now looking for participants experienced in mixing, mastering and live music production to score a number of music excerpts based on the amount of punch perceived. These scores will then be used to validate and improve our model.

The test will take around 15 minutes and can be found here:
https://selene.hud.ac.uk/u1364292/punch ... naire.html

Anyone who completes the test has the option to enter a prize draw to win a £50 (or equivalent value in the winner’s desired currency) Amazon gift voucher.

For those curious, the music excerpts used are from the Free Music Archive: https://freemusicarchive.org/

We would be very grateful for your help! And of course I’d love to hear your thoughts about punch.

Thank you

ShelLuser wrote: ↑

Fri Jan 10, 2020 4:26 pm

So the request for an e-mail address isn't optional? Then you can count me out.

So just to amuse myself I loaded up a proxy and such and went to the survey. I immediately got puzzled by the first question... so we need to determine if we're most experienced with mixing or mastering or "live music production" (what is that even suppose to mean?). The reason why I'm having immediate doubts when I read this is because mixing & mastering is normally an essential part of the music production process itself. So why you apparently split this and think you can treat it as several entities is way beyond me.

I'm also a bit puzzled why you think this is a good way to determine how users perceive this considering that a lot of the perception will depends on the users audio setup. Yet the survey never asks anything about that, which is where I drew the line and stopped taking this seriously.

Not to mention that your survey tried to load 3 (apparently huge) wav files, obviously failed and now it asks me to try again. Yah.... there are better and more reliable ways to do this, ever heard of SoundCloud?

.... in fact. I'm now starting to think that my anti-virus blocked all this though I'll get to the bottom of that. But if I'm right it most certainly would explain a thing or two... yah, this is why I use sandbox setups & a VPN first

The request for the email is at the end of the test. You can leave it blank and proceed as to not be included in the draw.

Live music production would be mixing music at a live music event through front of house, monitors, etc.

Splitting to mixing/mastering will be part of the analysis, looking at agreement between the groups and between the people within the groups to see if they can/should be merged.

The perception does depend on the setup to some extent, yes. In the test itself we're asking for headphone models used (the use of headphone is specified a the start of the test). We may exclude results depending on the answer to that, obviously something like apple earbuds won't make the cut.

As far as the test system itself, it doesn't perform as well as I would have liked either. However, it was the solution used given time constraints on the project. For some reason the test system is required to load the wavs prior to the test, that's just the functionality of the Web Audio Evaluation Toolbox.

You can rest assured it is certainly not a virus or anything close to one.

Thanks for your interest.

ShelLuser wrote: ↑

Fri Jan 10, 2020 5:53 pm

Thanks for clearing that up. I'm well aware that most people wouldn't think twice about this but I'm always a bit focused on privacy aspects, so I definitely appreciate the heads up.

Yes, I'm somewhat the same - but probably not quite on your level as far as VPNs and such, as much as I'd like to be. That's why I made it optional, because I know I wouldn't give an email out like that myself.

ShelLuser wrote: ↑

Fri Jan 10, 2020 5:53 pm

Apologies, I stand corrected on that part I have to admit that I totally glossed over that part of the initial description, in retrospect I think I started glossing once I read "it's recommended", figuring that I don't need to read that. Considering the small line of text I think it's easy to miss like this. But... still my bad.

I think your point is still valid though regrading perception changes on different systems. We'll have to be careful with the analysis and conclusions we draw, if any, from these results. It might be that the score distributions are huge and/or clustered - which is at least justification for further investigation.

ShelLuser wrote: ↑

Fri Jan 10, 2020 5:53 pm

No offense but I'm not taking your word on that

Take my word and my listening test! ...please

ShelLuser wrote: ↑

Fri Jan 10, 2020 5:53 pm

But in all fairness your response made this a lot more legit to me and I definitely stopped to stop taking this serious anymore (pardon the pun)

Yeah, one of my main concerns when going about trying to advertise this thing is how I can word things for people not to just immediately write me off as a spam bot, or anything else for that matter.

ShelLuser wrote: ↑

Fri Jan 10, 2020 5:53 pm

Thanks for the feedback! I'm going to try again later this evening because now I'm honestly curious about the music scores.

Likewise, I appreciate you taking the time to provide some thoughtful criticisms. Your participation in the test would also be very helpful. Thank you for taking the time.

For what it's worth your audio player system doesn't work in either Safari or Firefox on a Mac running Mojave. Safari endlessly fails to load the files and just loops while Firefox goes straight to the request for a name/email address. Which your site tells me mandatory and has to be provided to proceed.

I can't be bothered to make up an email address or use a throw-away, so sorry but it seems you really do need to get someone to take a long hard look at the ways in which the site is broken.

TLW wrote: ↑

Fri Jan 10, 2020 6:39 pm

For what it's worth your audio player system doesn't work in either Safari or Firefox on a Mac running Mojave. Safari endlessly fails to load the files and just loops while Firefox goes straight to the request for a name/email address. Which your site tells me mandatory and has to be provided to proceed.

Name question would be first and mandatory, but you can use any name you want.

I tested on Firefox and Chrome. Safari I looked at and it worked, but, I realise now this will have been before the Mojave update .

No worries though, thanks for the heads up, will look into this.

An update on this.

Last night I managed to test the system on MacOS High Sierra and Mojave on both Safari and Firefox without any issue. So for anyone reading through this, the above issue is somewhat isolated and won't necessarily effect everyone.

As for what the above issues are, I'm not 100% - I suspect the Safari issue might be more to do with the University servers being picky about the connection, which they can be, that's then dealt with poorly by the test system hence the looping. I'm assured by the University technicians that this should be a rare occurrence and is part of a known issue, but is frustratingly something I have no control over.

As for the Firefox issue, there's a couple of ways of interpreting:

One being, that the Name input box appeared, which is mandatory (though you can use any name you want) and this was mistaken for a request for email/address, in which case the system functioned as intended. If an input was given, there should have then been a request for headphone make/model which is also mandatory.

Second being, that the name input box appeared, a value was given, and on proceeding the optional email input box appeared immediately even though it's supposed to be given once the test has been completed. Then, the test couldn't be proceeded further until an email address had been provided. This is an issue that hasn't been seen before, and would mean the system is effectively skipping the whole test after the name input.

Sorry the system didn't function properly for you TLW, and thank you for taking the time to report this.

To add a little more info....

My MBP is still running Mojave still. My wife’s 11” Air is running Catalina. Both have the endless loop issue with Safari, though it has to be said Safari sometimes runs into problems with some streaming media such as Youtube. HDTV streamed from Now TV no problems, from the BBC iPlayer and website no problems. Youtube problems. iPad Safari on the other hand usually behaves itself with Youtube. My suspicion is that Safari has been made increasingly secure against an assortment of possible attacks and that has a knock-on effect on some media streaming platforms.

My wife’s Air runs virtualisation software (Citrix) to run WIndows server-side applications and databases in Safari without issues most of the time, though the last year has seen Citrix having to patch their applications at the server level to avoid Safari thinking it’s under attack in some way.

Firefox I’ll happily put down to user (as in me) error and misinterpreting the site’s behaviour.

TLW wrote: ↑

Sat Jan 11, 2020 8:46 pm

To add a little more info....

My MBP is still running Mojave still. My wife’s 11” Air is running Catalina. Both have the endless loop issue with Safari, though it has to be said Safari sometimes runs into problems with some streaming media such as Youtube. HDTV streamed from Now TV no problems, from the BBC iPlayer and website no problems. Youtube problems. iPad Safari on the other hand usually behaves itself with Youtube. My suspicion is that Safari has been made increasingly secure against an assortment of possible attacks and that has a knock-on effect on some media streaming platforms.

My wife’s Air runs virtualisation software (Citrix) to run WIndows server-side applications and databases in Safari without issues most of the time, though the last year has seen Citrix having to patch their applications at the server level to avoid Safari thinking it’s under attack in some way.

Firefox I’ll happily put down to user (as in me) error and misinterpreting the site’s behaviour.

Thanks for the extra info on this.

Your issue with Safari definitely seems like it's because of your connection to our servers, especially because you're seeing the same thing on both laptops, which I'm assuming were accessing the internet from the same network. Though why the same thing isn't happening on Firefox is uncertain. My own tests and the creation of the test system itself was certainly focused on use in FireFox and Chrome, so perhaps they are better able to deal with our server's connection behavior. Unfortunately ultimately there's not much I can do about this issue. It's an issue with our servers and the server technicians can't seem to get to the bottom of it because it occurs sporadically.

However, it's at least nice to have confirmation that what you experienced on Safari is part of a known issue (that I was told would be a rare occurrence. )

So, thank you for trying and for the information you've provided. I really appreciate you taking the time.

ShelLuser wrote: ↑

Sun Jan 12, 2020 2:36 am

Unfortunately I can't finish the test; you're asking way too much I think. I take time to listen to the samples and try to answer accordingly. That takes time. So after I had finished 10 examples I noticed that there were 'only' 100 more to go Sorry, but that's not going to happen for me.

This has been a difficult part. The issue being we need results over a large number of tracks to properly validate our model. We could have broken it into a smaller, more between subjects design, so each person only scores say 10 samples. But then there are a much lesser number of people scoring each sample, which we also need a lot of.

ShelLuser wrote: ↑

Sun Jan 12, 2020 2:36 am

I'm also honestly thinking that you're hurting your own project here. Concentrated listening takes effort, one way or another, but as always: if you try to ask too much from people then that will tarnish your results because people tend to get exhausted (extreme example) and most definitely 'saturated' over time.

Listner fatigue is definitely a concern. I think I should put in some intervals in the test where a message will tell people to take a short break, but again this will increase the test length. I will be looking at the time it took each listener to complete the test and potentially excluding people who took a long time on the basis of fatigue.

ShelLuser wrote: ↑

Sun Jan 12, 2020 2:36 am

But bottom line: having to go through 110 samples is way too much for me, sorry.

I completely understand. Thank you for your feedback.