Page 1 of 1

Har Bal reveals DAW sound differences

Posted: Mon Jul 26, 2010 9:16 pm
by Jetflow1
After changing speakers, applying room treatments, trying dozens of audio I/Os and various other things, I found I was still unhappy with the mixes coming from my DAW.

I decided to try one more thing; a test to see if each DAW software had a unique tone. I D/led the demo of Presonus Studio One, Magix Samplitude, Cubase 5. I also used the two DAW software titles I own; Sonar 8.5.3 PE (my current Daw) and Reaper. I chose these DAWs because the run on PC and they fall in my price range.

The test was done with Kontakt 4 running the same Abbey Roads 70's Drums presets in all the DAWs. All pan laws were checked for unity and all tracks were recorded as to the same level as possible, about -.1. All faders were set to zero and no busses were used. The same midi file was used in all DAWs (KitCore Terry Bozzio).

While there seemed to be subtle differnces between the various DAWs, to my surprise only Samplitude sounded very different. So I normalzed the mixes and ran them through Harbal and sure enough HarBal revealed what was heard to be true. Using Sonar as the reference file, you could see the differences. Cubase showed a bit more top end and a bit of change in the mid range and the bass was tighter. Reaper, Presonus Studio One and Sonar were almost identical with only minnor differences. Samplitude showed the most difference with the peaks at about 100 and 200 hz being smaller and narrower with much more mid energy than all the others. Har Bal REALLY helped nail this for me.

That was enough for me. I am now using the demo extensively to make sure it will work with my rig and to make sure it continues too sound good as I pile on the tracks!
Please understand this is not to say Samplitde is the best DAW. I just wanted to show how versatile Har Bal is. It helped me find the DAW that I feel sounds best to me.

Re: Har Bal reveals DAW sound differences

Posted: Sun Aug 08, 2010 2:01 pm
by Gordon Gidluck
I agree that Har-Bal is useful for telling the differences between two recordings!

But I am a bit surprised at your findings. Any professional DAW should give you identical results if recording the same source. This can be confirmed by recording the digital input and comparing the result to an original reference. I have done this with Samplitude many times in product testing my pocket pc recording application.

There must be some other factor. Simply normalizing the recording should not alter the tonal quality as this is just a multiplication of each sample by a constant.

I recommend going back and opening an identical sound file with all applications and compare. You should hear the same thing if there are no eq, plugins applied or any buss processing.

Re: Har Bal reveals DAW sound differences

Posted: Sun Aug 08, 2010 11:03 pm
by HarBal
The problem probably stems from how the experiment was constructed and the zero padding differences between the material being fed to the HarBal Analyser.

Quite a few people are under the mis-conception that spectrum analysis is more accurate than it really is. Averaging periodograms, which is what Har-Bal does, is a unbiased estimator of the true spectrum convolved with the effect of the data window applied to the time series being analysed. That being said, the estimate has variance (ie. uncertainty) and that variance component depends on the phasing of the time series with the analysis window.

To given an example, if you take a track A and put 1 second of silence on the end giving track B and take track A and put 1 second of silence at the beginning to give C, an analysis of B and C in HarBal will produce different spectrum estimates because the variance or noise component of the estimate will differ between the two. A naive inspection of the result would put the differences down to something in the processes producing the two files even though the common portion of the tracks are exactly the same. Personally, I think this is the basis of his difference and not something in the DAW itself.

To do a comparison like that requires the track layouts / timing to be identical. A similar issue arises if people wish to compare results they get with Har-Bal EQ to EQ with outboard gear because more often than not the layout of the two tracks being compared will not be identical owing to the user not knowing that it is necessary for a valid comparison.

This discussion raises the question of whether this presents a problem given that we are using HB analysis to make EQ decisions. That all boils down to a matter of degrees. When the EQ problems are of a larger scale than the uncertainty in the spectrum then you clearly can make a valid judgment. If the potential problems are small then it probably should be left alone, but in any case, we don't advocate flying deaf so given you listen to your EQ decisions and agree with them then there is no issue. As long as it indicates the trend accurately, which it does, then it is fine for the purpose. You could also argue that there may well be a whole lot more uncertainty about the spectrum if you just rely on ears alone.

cheers,


Paavo.

Re: Har Bal reveals DAW sound differences

Posted: Thu Aug 12, 2010 2:48 pm
by Jetflow1
Hey Guys,

First thing I want to clear up is how the test was done. I did not used a sound file I used a midi drum track firing Kontact 4 the drum samples I used was Abbey Roads 70's Drum Kit. The drums themselves were mixed within Kontact and saved as a preset so each and every DAW receieved the exact same midi file and the exact same drum preset. There were no submixes they went directly from the strip to the master fader and all pan laws were set to be identical. When I printed to test track I used no dithering and I made certain everything was set up the same. Therefore each and every Daw received the same source material.

The test was conducted as a hearing test to begin with only. I used two musician and two non-musicans and my wife. :) The tracks were listened to in my home studio, my friends studio and my two vehicles, as well 3 or 4 other sources. The songs were placed on several CD's in random order and only I knew which ones where which. The CD's were switched at my discretion. In fact during one listening test I actually switched CD's during one of the passes and to my amazement when the results were tallied every person picked Samplitude every single time. No one was able to discuss the results until all of their answers were turned over to me to analyze. Har-Bal was only used as a visual test and there where differenence. Don't get me wrong the differences between the DAW's were very slight except for Samplitude and Cubase. Cubase was marjoanily different but Samplitude was very different, so different in fact that Samplitude is my new Digital Software Package. The best way to try this out is to try Samplitude you will not believe the differences until you hear it. I did not used Har-Bal as the desiding factor I used about 14 ears to help me decide Har-Bal was just confirmed what I was hearing that all.

Also I want to point out the reason I did the test using VST's and midi files is because that how I work. I do not use audio files what so ever nor do I use any out board gear. Everything I do is completely in the box. You may get different results if you use audio files instead of VST's


Thanks,

Mike

Re: Har Bal reveals DAW sound differences

Posted: Thu Aug 12, 2010 8:26 pm
by HarBal
Given that the source was a VST device it may be that the differences between the DAW's you tested have to do with the implementation of VST hosting. A VST plugin can connect to the host with a number of difference data representations (32 bit float, 64 bit float, 32 bit fixed point etc) and different process buffer sizes. Which is used will depend on the protocol exchange between DAW and plugin and what the DAW natively supports.

If different formats get used in different DAW's then the data coming from the VST plugin will have a different data path and those differences in sound might simply be down to differences in how the plugin handles different formats, rather than something inherent in the DAW. It is possible that the sound differences are down to the plugin and not the DAW. That is just the nature of host plugin interaction.

The fact that it works best with Samplitude may simply be because they share a common internal representation. That being said, I have a friend who swears by Samplitude. I don't doubt what you are hearing, I just think a deeper look is required to understand where it is coming from and I doubt it is inherent in the DAW itself, but rather a byproduct of how the DAW and plugin interact resulting in different processes being used.

cheers,


Paavo.

Re: Har Bal reveals DAW sound differences

Posted: Sat Aug 14, 2010 1:36 pm
by Gordon Gidluck
Mike,
I understood how the source was generated. I was suggesting using a rendered file for comparison purposes to keep things as controlled as possible for your experiment.

Coincidently I have been a long time user of Samplitude. I would also add that it could be differences in their mix engine. Certainly they have had lots of time to perfect it. I can relate what was posted on the old Samplitude newsgroup (precursor to the current forum.) Interestingly, how they started out was with mixing samples on the Amiga. And they always have used 32-bit float for internals. Eventually one of them had the idea to start developing a recording application that eventually became Samplitude. But from the beginning, mixing samples had been the focus.

However, I am a bit surprised that Reaper did not fare better in your test, as it uses a 64-bit data path throughout. I would think that it would handle VST plugins in any format nicely. Maybe Samplitude just works optimally with your setup(?)

I totally get your point that Har-Bal helped you choose Samplitude. I am just trying to figure out why Samplitude came out on top of this test.

Gordon

Re: Har Bal reveals DAW sound differences

Posted: Fri Aug 20, 2010 8:20 pm
by Jetflow1
Hey Gordon,

My main focus was to eliminate this sound in my mixes that was driving me crazy. At first I thought it may have been the room, so that is where the idea of arc came. But to lay out that much money on a hunch did not make sense so I got the DBX unit and ran it. Instead of using the first reading to set the EQ, I used it to adjust and add some of the bass traps. I was able to make the setting very minor after a couple of days and test and tune..but the sound was still there. The sound was most noticable in my abbey road drums . I did not like the sound of them at all and after days of compressing and eqing to try and make it sound realistic, I started looking for another solution.

I own Reaper, so I pulled it up and tried the AR drums there. The sound was was pretty close to Sonar. So I dug out a copy of Cubase LE that came with my E-MU 1212m card and loaded it up.Then I Googled "best sounding DAW" and up came Samplitude. I D/Led the demo as well as the new presonus Studio One demo and that is when the the "experiment" started.

Now the problem is solved! My AR drums sound fantastic and all my VSTs sound better than ever. I am just about out of time on the demo and I will be picking up my copy in the next couple of weeks. The mixes sound great too.

Now there are a few things I need to say here about this whole thing:

- This was NOT, by any means, a scientific experiement. This was just my way of trying to eleiminate a tone I did not like.

- I am not stating that Samplitude is the "best soundung" DAW software. What I am saying is that for my purpose, my room, my gear and my personal taste, I prefer the sound of Samplitude over Sonar, my current DAW software.. Being 49 this year, I am well rooted in the analog era and the sound I was trying to rid myself of was what I come to refer as "digital aftertaste".

This is my first DAW as well. I have used computers since 1988 but only for sequencing, never sound source. I used tape until this.

-I am also not stating that the other DAW software sounds bad either, just different. I started listening to more modern recordings and realized that this sound was there as well and that my AR drums mixes on Sonar sounded VERY similar. I guess I have been listening to too much 70's Prog rock....if one could say that there is such a thing as too much 70's prog rock. :)