danbronson

I don't know if this is valid, perhaps someone here could comment, but I've thought about hanging carpet a bit out from the wall so the sound wound be dampened a bit as it passed thru the carpet before reflecting off the wall, then attenuated a bit more as the reflected sound passed thru the carpet again. As mentioned above, this would cut down on the reflections but not provide proper 'treatment' Yes, this is the best way to do things. With carpet, an inch or two should be fine. The result is an extended (lower) frequency range that can be absorbed. It'll be more significant with thick bass traps though. One other thing to mention... If you're going to hang carpets, make them the thickest carpets you can find. If you have a lot of them, rolling a few up and placing them in the corners of the room can make for a decent on-the-fly bass trap. It is usually more important to deaden the lower frequency range than the highs. High end resonance in a room just sounds like reverb. Low end resonance can create huge comb filtering problems.

Hanging a carpet or two will definitely deaden things in the higher frequencies (which may or may not be a good thing, you'll have to figure that out for yourself by experimenting because every room is different). The result will sound less 'live', though any low end and midrange problems (which any room in a typical house will have) will not be fixed. The uncontrolled low end will cause monitoring problems mainly, though your recordings will suffer if you set up room mics too. Low end/midrange issues can be treated somewhat effectively with bass traps, though that is somewhat expensive and/or time consuming (I have made a few) and they rarely solve a problem completely. To really stop these problems, you need a room designed with acoustics in mind. Acoustics are one of the most important aspects of recording (all the rack equipment in the world won't help if you record/mix in a shitty room) and one of the most difficult/expensive to get right. Not trying to sway you from setting up a home studio though. Looks like you already have realistic expectations.

Like I said, the change in transient response, smearing and (what sounds to me like) low pass filter are what is making the clips sound different on a whole. That's what I'm hearing, not performance differences. I paid extremely close attention to how I performed each test to get my performances as alike as possible and the mic positioning has not changed. If my playing were inconsistent, you would hear it within each clip. Even if inconsistent playing could make the changes you hear, it would be very unlikely to affect each entire clip in a completely different, yet consistent manner. For you to assume that it has downright stupid, hence my frustration. Go ahead and reread that paragraph until you understand every word, because it's certainly not getting through to many people here. So there's that, but there's also the fact that I performed three separate tests (which you've heard two of, the first was deleted because I wanted to do it again to verify what I heard the first time, therefore minimizing the possibility of error), all with the same results. I also recorded a pass at 96 kHz and added it to the guitar clip, just before the 48 kHz pass (so it is first). You can hear that they are very similar sounding (96 sounds very slightly better than the 48 to me, just my opinion), while the 44.1 kHz clip without r8brain conversion sounds dull, muddy and lifeless. http://dl.getdropbox.com/u/1395881/96%20then%2048%20then%2044.wav So that's seven seperate performances in total (4 on the cymbal, 3 on guitar). And every time, the recordings that started at 44.1 sound dull and crappy compared to recordings with higher sample rates. You'll have to take my word for it on the cymbal clip you didn't hear. I deleted it because I wanted to use the same file names for the second test. Needless to say, the 44.1 clip suffered from the same dull sound. Obviously we're dealing with two variables here (performance and sample rate) and it's not possible to isolate them. But if the test is repeated more times with the same results, the accuracy of the findings goes up. The only problem is, I'm not going to repeat this test a hundred times to satisfy you guys. Playing this stuff consistently, seven times is enough for me. So let's leave it where it is. Nothing more needs to be said. If you want to argue more, read through the posts I've already made because I've said all I need to say. Make up your own mind about the findings I've posted here. /last post in thread

If your built-in SRC algorithm truly sounded bad, then I promise you that it would impossible to improve it by running an additional SRC algorithm on top of it. This is assuming I accept your theory about AD conversion requiring a built-in SRC algorithm, which I do not. I do appreciate your post, though. I just don't agree with the assumptions you made going into your test. They're not assumptions. Do a little research. I'm not going to spend an hour or two reading and quoting just to respond to you accurately (even if I'm generally right, I would probably miss a detail or make a generalization at some point that would open up a whole new bag of {censored}) so feel free to do it yourself. Or, if you're certain I'm wrong, feel free to explain why. EDIT: Looked into it a bit now that I have time, and I've been misinformed. I was under the impression that all converters worked like DSD audio (1 bit/2.8224 MHz). But all that is really beside the point, which is that regardless of the way you get to 44.1 kHz, there will be filters applied to the audio that can potentially sound terrible. Therefore, two different ways of getting to 44.1 kHz can potentially sound different. And in this case, they do.

I'm in a bit of a rush right now so I can't go into detail or double check what I'm about to say but... When an A/D converter puts out a digital signal, it doesn't start as 44.1 kHz or 48, 96, etc. I think (I could be wrong) but it starts out as a 1 bit file with an extremely high sample rate. The waveform is represented as accurately as possible in an extremely primitive form (1 bit = just 1s and 0s, on and off). The A/D converter isn't done yet though. It uses a built in sample rate converter (the same thing you'll find in software, like r8brain) to send the audio to the computer at the desired sample rate. The built in sample rate converter can potentially sound terrible. That's why it can be beneficial to record at a higher sample rate (where there will be more allowable detail and the filters will be much further outside the range of human hearing). In my case, it is beneficial. I encourage everyone to check with their own converters. Voxengo r8brain is free and very high quality. Do a test like the one I did and listen for yourself. If you don't hear a difference, then the built in sample rate conversion in your A/D is good enough for you. If you do hear a difference, use the method that sounds better to you.

I hope you can pardon a perhaps overly broad analogy but that seems like trying to choose a paring knife by comparing paring knife A cutting an orange with paring knife B slicing an apple. I recognize you're not trying to do a scientific test, but I'm afraid I just don't see the point of a comparison where you cannot isolate the factors that you're setting out to test. Feel free to propose a way to do this. I would love to remove that one last variable. Until then, it is what it is. I feel there's a lot to be learned from this comparison. But maybe I'm the only one who bothered to even listen to it. Perhaps next time I will just keep my findings to myself unless I happen to run them by CERN first. I only posted so that people on the forum could learn what I have just learned. But if you guys want to waste your time chasing impossible testing standards, be my guest. I'll be over here improving my ears and recording skills in the mean time (starting with using higher sample rates on the way in and converting later).

So basically, this is not a sample rate conversion shootout. It very much is. It compares: -TC Studio Konnekt 48's conversion to 44.1 kHz -TC Studio Konnekt 48's conversion to 48 kHz + Voxengo r8brain's conversion to 44.1 kHz Note that it is not possible to compare the r8brain on it's own to the TC SK48, as the SK48 must do some conversion if it's doing A/D conversion. Therefore the best way to test is to use the SK48 at a higher sample rate (which will be more transparent) and use something else to bring it back down to 44.1 kHz. Put plainly, this is a comparison of two different ways to get audio to 44.1 kHz using the same A/D converters.

I'm not arguing with your opinion, I'm just curious why a different sample rate would affect transient response. I'll listen to the clip tonight and give my opinion. Yeah I didn't think it would be able to do that either, but check out the link blue2blue posted. It becomes a little more obvious how transient response can be effected by sample rate conversion.

Call me lucky I guess, I was taught scientific method in school. I'd use it if I thought something like this was important enough (not that I can anyways), but I guess I don't take myself that seriously. It's just a sample rate conversion comparison after all. I stand by what I said about a good ear being able to discern what sounds different due to performance differences, and what sounds different because it's making an entire take sound cloudy and dull. :/

At the risk of offending, that is such a frustrating cop out of an answer. :/ A good ear (with good monitoring) should be able to hear which differences exist due to performance, and which exist due to sound quality differences. For example, if a low pass filter were applied to one clip at 5 kHz, the lack of top end in just one clip would obviously not be caused by the fact that one is a different performance. This situation is not far from that example. Are the performances different? Of course. Are those differences going to make one clip's overall spectral balance and clarity noticeably different than the other? Hell no. Don't kid yourself. Actually listen to these things, the way they differ could never come down to performance... To me, the differences in these clips are not subtle. I'm surprised nobody has said anything. Even my girlfriend and parents reported almost the exact same findings in blind tests (and with no communication with each other). And frankly, I agreed with every word they said. No cop out answers, they just said what they heard. What I'm hearing is this (since nobody else has the guts to voice an opinion )... The first halves of the clips sound good to me. The top end is quite apparent, making things perhaps a bit thin sounding. Nevertheless, things sound clear and punchy to me. I listen and feel like the sound is actually happening right in front of me. Cool. The second halves of the clips sound very, very dull and 'smeared' in comparison. The transient response is basically gone, making it seem like the sustain to attack ratio is higher and therefore things lose a lot of punch and also sound more compressed. Also, it sounds like there is a low pass filter on the top end (and indeed, there probably is since the differences here are in the sample rate conversion method). The lack of high end means the lows/mids seem much louder in the second halves, and it's here that the 'smearing' sound is so obvious. There is just an overall lack of detail in the whole frequency range when compared to the first halves of clips. It all sounds like dead, muddy {censored} comparatively! For what it's worth, I did the cymbal test twice (but only posted the second one) to make sure that the results would be consistent between both tests (and of course, further tested with an acoustic guitar for three tests in total). What can I say? Exact same results every time. So, since nobody's playing...the first halves (bright and lively, remember) are recorded at 24 bit/48 kHz, exported from Cubase 5 (24 bit/48 kHz) and converted to 24 bit/44.1 kHz with Voxengo r8brain free. The second halves (dull and dead) are simply recorded at 24 bit/44.1 kHz in the first place and exported that way as well. Then they are all brought into the same 24 bit/44.1 kHz Cubase project where they are level matched and the volumes raised (without gain reduction) in Voxengo Elephant on the master buss. They are also dithered to 16 bits with Elephant and exported as a 16 bit/44.1 kHz wave file. This sound quality difference is a pretty huge revelation for me, as it means I can get a much better clarity in my recordings by recording at a higher sample rate and converting later with r8brain.

Here's a handy SRC comparator: http://src.infinitewave.ca/ It takes a second to figure out the interface -- it's set up to select one SRC on top and a different one on the bottom and then compare aspects of their tested performance. I can't vouch for the rigor or methdology but they've been updating this thing regularly for some years now and no reason to not think it's as informative as it looks. Funny you'd post that, this site was the inspiration for these tests! I guess people either aren't hearing a difference in these clips or are too worried about being 'wrong' (not that there is a 'wrong' here) to post, so I'll spill the beans on what the differences are. Until now, I've always just recorded at 44.1 kHz and left it at that. The files are smaller and they don't have to be sample rate converted to become CD quality, just bit truncated. I figured this was saving me some audio quality loss but I guess I never really thought it through till now, because even if I'm not converting in software, my A/D conversion is still sample rate converting to 44.1 kHz! So you can't really get around it. Either way, you're sample rate converting at some point. Knowing this and seeing how differently software can test from the site you just posted got me wondering... I changed Cubase (and therefore my converters) to record at 48 kHz. Then I downloaded Voxengo r8brain (which tests extremely well on that site) and used that to convert to 44.1 kHz. I compared that to a clip just recorded at 44.1 kHz as I have always done before. That's what's going on in the clips I posted. Can you hear it? Which one do you think is which?

The biggest difference is the performance. It is definitely unfortunate that I couldn't use the same performance somehow. But I'll say this, they're not as different as you might think. I found that one method of recording completely changed the transient response (or at least that's what my opinion is, I want to hear everyone else's!), making one seem as if I were hitting harder and/or in a different place on the cymbal. But in truth the performances are very, very similar. I guess you'll have to take my word for it though. Still interested to hear what people's opinions are. It's not a trick or anything, I promise! Just let me know how you think the clips compare. It'd be nice to get some discussion going here, even if people disagree with each other (especially if they do!). Here is another clip. The order (concerning the way these were recorded) is the same as in the first clip. This time it's two acoustic guitar clips (about 6 seconds each). The two clips have been RMS level matched to within 0.01 dB of each other. This'll show how an instrument with a wider range reacts to the two methods of recording (I will explain them once I get more opinions, but I don't want to bias people and I've already said too much). Hopefully this makes it more obvious than the cymbal? http://dl.getdropbox.com/u/1395881/SRC%20Shootout%20Acstc%20Gtr.wav

Thanks for listening. Anybody else?

Hey, I have a 24 second long CD quality wave file here (it has been dithered from 24 to 16 bits with Voxengo Elephant, no limiting however). There are two different performances on a ride cymbal (first at 0:00, second at 0:12) that were recorded/sample rate converted in different ways. Their RMSs have been matched to within 1 dB of each other. That's all I'll say for now. http://dl.getdropbox.com/u/1395881/SRC%20Shootout.wav I want to know what you think of the differences. I've already made up my mind, but since I can't do a blind shootout on myself, I want your opinions as well. Are there big differences? What are they? Which half of the clip do you prefer? Maybe the differences are smaller or not noticeable to you? Please let me know! Thanks!

OH NOES!1 quick! tell him he's a stupid head!