Critical Listening, Any Tips?

LOUT

I've heard that certain songs and noise/tests can be good at revealing particular flaws in a design or driver or crossover. I've seen some recommended listening lists, but I haven't noticed much mention about what specifically to listen for problem-wise and I'd like some guidance on other listening test using noise or songs or whatever else. 

Are there some songs or clips that are good at revealing frequency peaks/valleys or generally unnatural frequency response?

Is there anything I can listen to that'll be really good at revealing distortions?

Is playing pink noise while moving around and trying to listen for obvious "shhweeiiiiw whoosh" of comb-filtering and notches a good way to "test" for phase/alignment issues, or are there more effective listening tests?

Are there any other major speaker problems that can be heard and narrowed down through some careful listening tests?

What are some important pitfalls to lookout for when trying this?



I think I'd like to become a more effective listener with a better ear (though I'm also a little afraid that ignorance is bliss). Teach me your ways. Empower me to put off buying measuring equipment and learning to use it for as long as possible.

R-Carpenter

Listening to live music always put's reproduced sound in a good perspective.

Also having a preferably "reference" level speaker for comparison is a good thing. What I mean by "reference" is very linear on axis frequency response and known off axis response.

comb-filtering is audible as you move your head in a particular spot. You would hear disappearance of a range of frequencies. Line arrays show this behavior (how drastic depends on the design)

Speakers generally are a set of problems that you trying to balance and alleviate.

LOUT

I think I don't know how to use this forum very well, lol. I don't want to mark this as answered because there are several questions and I'd like to hear multiple opinions, but "rejecting" the answers in the post above feels rude. 

I also think marking this as related instead of DIY might've been a bad choice both because it may get less DIY attention and because it's technically for trying to help choose between drivers and tune crossovers. Not sure if that's related to DIY or direct DIY.
I'll see if I can change the format to something a little more accurate and maybe able to get more attention.

LOUT

Nevermind. I could be blind, but I'm not seeing the option to edit existing posts. 

rjj45

@LOUT - as you build your experience in the hobby you will develop your own "go to" set of recordings that can reveal certain problems in speakers. We all have our preferred albums and tracks.
For instance, I know that Bonnie Raitt has a tendency to be a little bit sibilant on her "esses" - so if I hear an objectionable level of that, I'll tweak it down a bit. Similarly, you will find some bass torture test tracks. 
There is a long thread on Parts Express Tech Talk for suggested tracks. I don't agree (or know) many of them, but that's a good starting point for exploration - you probably have a few of the albums in your personal collection.
Hope this helps.

rjj45

re: comb filtering
I once had a set of EPI (I think) speakers that used 2 midranges side by side.
The sound changed drastically if you moved from left to right about a foot.
That's comb filtering!
I have not noticed a severe problem like that on any other speaker.

jr@mac

Personally, I have a number of tracks I use snippets from. Some are used for sibilance sanity checks, other for baffle step correction confirmation. There are tracks I use to get a feel for imaging, dynamics, etc. These tracks range across a variety of genres. I have about 7 minutes worth that I downmixed into a single mono track, as well as original stereo. They are level matched. I play the mono track over and over during design phase, and will use the stereo track as confirmation. I have not had to make changes in many, many years after the stereo confirmation. 

This is not critical listening in the classical sense, this is to help ensure when I sit down for listening - I know I have the bugs more or less worked out and I am less prone to wasting a glass (or three) of Scotch and an hour or two listening for shortcomings. Sometimes if I am listening to music I am unfamiliar with, my brain will try to convince my ears I missed something, but I have become pretty good at telling my brain to shut the hell up and enjoy. 

I will not be sharing my test tracks, not out of secrecy or anything like that; more because they just are not that interesting. What is important is to use music you are familiar with, and listen to a new design against a reference that you find pleasing. 

dcibel

@LOUT For most of what you are asking about, a mic is the best tool. Use it with the combination of your ears to make correlations between what you hear and what you measure will help you understand what's going on a lot better and make educated and specific changes to your system.

As for critical listening music, I highly recommend just listening to music you like, songs that you've heard 100 times before that you know what they should sound like. Mixing and mastering play a big role in sound reproduction, so trying to set up your system with an unfamiliar track list is not something I'd recommend.

LOUT

One of the reasons I'd like to edit my first post is to change the "comb filtering" part to simply say "timing/alignment and phase issues"...even though they're technically versions of the same thing, I think a lot of folks tend to recognize combing as a much more drastic and noticeable problem while more subtle or smaller phase/timing/alignment issues are a bigger part of what I'm wondering about because I'm not sure what I could be missing or generally ignorant of.

My current DIY 2ways seem to have what should be a somewhat obvious deviation from what modelling shows where reversing tweeter polarity to the woofer should be more ideal but I literally could not hear a difference in practice even while A/B-ing between them with one wired -+/-+ and the other -+/+-... That includes music and sine waves (sadly I didn't realize the potential of pinknoise at the time). 
I'm guessing this is likely because both my woofer and tweeter are simply mounted on-surface rather than flush-mounted or offset forward/rearward. I have more limited experience looking at the effects of this alignment difference in modelling, but accounting for the difference still makes it look like the difference between wiring the tweeter with matching polarity or reverse with the woofer should create a noticeable difference in frequency response (or at least some kind of phasing issue when moving around). 
It's bugging me wondering if this is the kind of thing others might hear as obvious while I'm totally oblivious.

In short, I don't want to blindly make any embarrassingly bad mistakes that are within my likely considerable blind-spot while others give my creations a listen and think "eww, what a deaf noobie" despite being more polite externally. So I'd like to narrow my blind spots as fast as possible. Preferably before MWAF. 


Also, once again please don't be offended that I'll mark your answers as not-answers even though they are valid and helpful. I'm just assuming that leaving the post marked as "answered" will mean other folks with advice may ignore it. 

NavyGuy

I wrote an article on our website with the test tracks I use: https://www.css-audio.com/single-post/2018/02/07/Test-Tracks-for-Evaluating-Loudspeakers

dcibel

Spend the $75 on a calibrated mic, UMIK-1 or UMM-6, you won't regret it. For a simple plug and play solution with absolute SPL capabilities, consider the Omnimic.

Kornbread

Here are a few track I frequent when trying to figure out what I've got: 1) Cowboy Junkies, Sweet Jane - The s'ss are hot.  If everything is right in that area the s'ss will be hot to the point of almost being over-the-top.  If it's over-the-top, pull that area down.  Lorde, Liability - The S'ss can stick out some but any more than that is too much and her voice should be pinpoint dead center and several feet behind the speakers.  2) Natalie Merchant, Peppery Man - Everyone should be played out on stage exactly as they should be. Roger Waters, Amused To Death (the entire album) Sounds will emanate from everywhere, also some good dynamic swings.  Chris Jones, Long After You're Gone - Should be able to see him sitting on a stool picking those strings along with tapping his hand on the guitar or maybe it's his foot on a wood floor.  3) Pentatonix, Daft Punk, The bass line is a bit overboard.  If the speakers are too generous there, the bass line will overload the entire song and it sounds like a blob.   

                

rjj45

Depending your exact crossover type and topology, you may not hear or even measure any distinct differences if/when you reverse polarity on the tweeter. A very common topology is 2nd order on the woofer, and 3rd order on the tweeter.
When you get the phase in the crossover region matched, you will see a deep notch at the crossover frequency when you reverse the tweeter. Somewhat magically, the imaging will suddenly become very clear, as well as subtle musical aspects that you never heard before on your DIY projects.

LOUT

NavyGuy said:

I wrote an article on our website with the test tracks I use: https://www.css-audio.com/single-post/2018/02/07/Test-Tracks-for-Evaluating-Loudspeakers

Thanks for the link. I found a similar list like this on a different site a bit ago, but I feel like this particular one gives a bit more detail on what to listen for specifically which is quite helpful. I understand the suggestions to listen to familiar music, but that doesn't help if I'm only familiar with it when listening on other heavily colored speakers/headphones and those tinted sources are my main "references". Playing around with an EQ has taught me that I can have a pretty forgiving ear for many inaccuracies at least after a short break-in period at least as far as considering things comfortably listenable and "nice enough". I agree that a good reference speaker would be pretty ideal for listening comparisons, but I don't have any currently and the cost would likely rival a $75 measuring mic which I'm cheap-skatingly trying to avoid for now.

dcibel said:

Spend the $75 on a calibrated mic, UMIK-1 or UMM-6, you won't regret it. For a simple plug and play solution with absolute SPL capabilities, consider the Omnimic.

I'm glad to hear the prices aren't outrageous and I'll probably go this route eventually if I stick with this long enough. I should probably just bite the bullet now since a lot of this measuring and tweaking is often something I enjoy about as much as the actual listening..but I'm stubbornly trying to do this by ear for as long as possible for admittedly no really good reason. Sorry, I understand it can be frustrating to give someone advice only to have them basically ignore it at their own detriment. On a related note, how far off do you think a "flat response"-claiming, large-diaphram, recording condensor mic should be if some free software even accepts it as an input device?

Kornbread

As far as our ears/brains becoming used to the sound; If listening to a new creation and I can listen for a long time and find myself relaxed, or even asleep vs getting up to turn the music off. The speakers may not be quite there yet, but they are getting there.        

GeoffMillar

I second the suggestion to check out the PETT discussion on recommended test tracks, there's some good discussion in there.

I'll make a few comments, none of which are new, of course:

you should listen to a range of music you know and love, and it should be of varying degrees of recording and mixing quality. To trot out my pet peeve, if the Red Hot Chilli Peppers 'Californication' sounds good on a speaker, nothing much else will as the album is soooo compressed, limited and shrill.  I wouldn't test speakers with music I don't like. My brain would be saying 'turn it off'!

I use a range of music for testing: classical chamber music, well and poorly recorded rock/pop, jazz and operatic recordings.  Probably 20% of our collection consists of old/poorly mixed or bootleg recordings, so it's important to be able to enjoy that music.

Spoken word/comedy is also a good way to test aspects of bass performance. For example, one of our newsreaders is nick-named "three balls" due to his deep and resonant voice and if a speaker has exaggerated or poor bass response he will sound awful.

I don't have measuring equipment, but I'm happy as long as a range of music sounds good.  Of the speakers I've built- all other peoples' designs - only two have what you'd call a 'flat response', but they all sound good to me.

Geoff

dcibel

LOUT said:

On a related note, how far off do you think a "flat response"-claiming, large-diaphram, recording condensor mic should be if some free software even accepts it as an input device?

It may be flat through the mid range, but you very likely won't have accuracy in the top octave(s). Depending on what you want to do with it, that may or may not be a problem. Correlating "changes" doesn't necessitate a calibrated mic, as the relative difference between two measurements will still be there.

Tom_S

LOUT said:

 On a related note, how far off do you think a "flat response"-claiming, large-diaphram, recording condensor mic should be if some free software even accepts it as an input device?

If you have a USB/FireWire/PCI audio interface with a decent mic pre, you can use any mic with ARTA or Holm Impulse - probably REW too, but I've never used that one. Most budget LDC mics that I've seen are not quite as flat as a calibrated measurement mic. But like dcible said - you're just looking for the differences between the two, not necessarily a flat line.

LOUT

I'm assuming an anologue interface (mic-pre output and PC 1/8"mic-intput) doesn't work with most measuring software?
Decent accuracy from roughly 500hz-5000hz would be plenty. Is that a reasonable expectation for an old fashion XLR-to-micpre LDC in the $150-$200 range?...if a lack of digital pre or USB mic isn't a dealbreaker already.