DP vs Nuendo

[kwiz]

It has a Nuendo 2 example, but not Nuendo 3.
I wonder if they have the same audio engine? If so, it would would explain why Timeline is hearing more top end in DP. The whole subject is very interesting.

<small>[ August 16, 2005, 07:45 PM: Message edited by: kwiz ]</small>

[grumph]

The "secret page" only talks about the way it sounds when you change the sample rate of a soundbite.Not about the way the DAW reads the audio.
I'm sure you can hear differences when you play a song with different softwares.
By the way,after a lot of testing,i've found that the "old" Spark was the best to change the sample rate.Maybe that's why Audio Ease do not compare Barbabatch to Spark...

[Timeline]

I don't think this has anything to do with it. All of my audio was 96K and no down-sampling occurred.

But.... SDII was converted to AIF and that could be the problem.

I asked if this was a possibility and those supposedly in the know said it was not. I find that hard to believe but I know where they were coming from so maybe it's something else like the way the PCI424 card dealt with audio via Nuendo 3.

I don't know.

I need to try some recording i guess but my projects are all done.

added: the engine is the same on 2 and 3

<small>[ August 16, 2005, 03:26 PM: Message edited by: Timeline ]</small>

[qo]

Originally posted by kwiz:
The main thing that impressed me was the sound. They had a PT HD system which I've worked on before but I've never heard PT converters sound that good. When I asked my friend (who owns the studio) what he did to improve the detail and imaging that I was now hearing he told me nothing.
What I was hearing was how good Nuendo's audio engine was.

What's your friend's studio like acoustically? Perhaps your friend would let you bring DP over to A/B in the same space? In other words, maybe the space has been well engineered/tuned and that's what you're hearing?

[mhschmieder]

Kwiz says Nuendo sounds better; Timeline says DP sounds better. Timeline's response seems to be intended as reinforcement of what Kwiz said. I'm definitely still confused!

[Timeline]

Well Mark, Remember I was importing mixes from DP which were SDII. I have not tried recording in Nuendo yet.

With my limited attempts to recreate mixes from DP I have found sonic problems with Nuendo via OMF import from DP etc.

I have also read posts of users claiming they dislike the conversion sound of Nuendo.

I have seen NO posts of anyone having a problem with the recording "sound" of Nuendo.

I have also read posts from users asking about ASIO configurations on how to improve sound on import to Nuendo

Hope this helps.

<small>[ August 16, 2005, 06:47 PM: Message edited by: Timeline ]</small>

[kwiz]

Originally posted by qo:

Originally posted by kwiz:
The main thing that impressed me was the sound. They had a PT HD system which I've worked on before but I've never heard PT converters sound that good. When I asked my friend (who owns the studio) what he did to improve the detail and imaging that I was now hearing he told me nothing.
What I was hearing was how good Nuendo's audio engine was.

What's your friend's studio like acoustically? Perhaps your friend would let you bring DP over to A/B in the same space? In other words, maybe the space has been well engineered/tuned and that's what you're hearing?

I hear you but the room acoustics didn't change. I've done sessions out of there using his HD system with DP as the front end. I noticed the difference when he was tracking/mixing a session using Nuendo. The improvement in sound wasn't mind blowing but it was noticable. Again, I guess there is something to be said about having better algorithms in the audio engine.
DP for me is still number one. My Rosetta 800, Big Ben, DAC-1 and the newly added 2bus LT has my friend wishing he never bought PT HD in the first place. He wants to try out Nuendo on my system to see if the "sound" of Nuendo can be improved further.

[qo]

Originally posted by kwiz:
I've done sessions out of there using his HD system with DP as the front end. I noticed the difference when he was tracking/mixing a session using Nuendo. The improvement in sound wasn't mind blowing but it was noticable. Again, I guess there is something to be said about having better algorithms in the audio engine.
DP for me is still number one. My Rosetta 800, Big Ben, DAC-1 and the newly added 2bus LT has my friend wishing he never bought PT HD in the first place. He wants to try out Nuendo on my system to see if the "sound" of Nuendo can be improved further.

Cool. I'd be really interested in hearing how it goes if he does bring Nuendo over to your place kwiz. If we can say with some assurance that there really is a difference in sound quality between these DAWs, then we can start banging on MOTU for something better

[Shooshie]

Well, in my non-engineer's opinion, what those charts (on the secret page) really showed us is each DAW's handling of post-Nyquist frequencies. All audio has to be anti-aliased at some point, doesn't it? Otherwise you get various phase-reinforcement effects that give digital audio its trademark nasties. A pure track coming in from your audio interface's A/D converter has already been filtered and dealt with by the A/D converter at the sampling frequency, but as soon as you combine any two tracks together or add a plugin, which all DAWs do, you've got a problem. Thus, we have to dither. Wouldn't there be a direct relationship between dithering and rate-conversion anti-aliasing algorithms?

[EDIT: I got to thinking about what I said, and realized that Dither has more to do with bit-depth change, rather than general mixing of audio. I guess what I said here is probably all washed up, but I'll leave it anyway, because it may trigger someone with real engineering knowledge to explain the real implications of those charts.]

Of course, I'm guessing. While there COULD be a relationship between the two, it does not mean the programmers made that choice. They could use an altogether different algorithm for dithering. But it seems to me that the terrible filtering exemplified by Nuendo's sample-rate conversion would be indicative of similar problems with audio in general.

In fact, if you look at DP's chart, I think that it very clearly reinforces what my ears tell me when I record with 48k sample rate, as opposed to 44.1. The difference between 44.1k and 48k, for me, is night/day. 48k opens up the sound spatially and makes the audio clear and beautiful. 44.1k suffers from a "stuffiness" in DP which I've heard many people complain about. Well, that could very well be due to the problems inherent in their anti-aliasing filters beginning their cut-off slope at about 18KHz, as opposed to 22K.

How many of you have heard what I'm talking about? Wouldn't this explain it? Wouldn't this imply a relationship between rate conversion and general audio handling? We're looking at the magic bag of tricks here: what to do with frequencies you can't hear, which has a huge effect on the frequencies you can hear. And while neither DP nor Nuendo performed great at 44.1K, according to the chart I would expect DP to dramatically outperform Nuendo at 48K, which is where I record.

Also, it looks to me that it would be better to use Barba-Batch or Spark to drop down to 44.1k for mastering to CD, as opposed to using any DAW for that purpose.

Shooshie

<small>[ August 16, 2005, 11:13 PM: Message edited by: Shooshie ]</small>

[buddhabelly]

Originally posted by Timeline:
Well Mark, Remember I was importing mixes from DP which were SDII. I have not tried recording in Nuendo yet.

There is no difference in quality between SDII and aif. They're both uncompressed audio file containers. Why would you think there would be a difference of sound quality between the two??

IMHO, and I'm by no means an expert, it is impossible to compare DAW like this. Each engine, with the exception of ProTools is 32-bit float point resolution. 24-bits of which are used for the fixed resolution of the sample and the remaining 8-bits are used for scaling.

The only DAW that you can really compare bit depth vs. sound quality to is PT, which as 24-bit processing into AND out of each plugin, thus causing a slight loss of information, but is most likely as negligible as the normal errors that occur in the math when calculating floating point math.

IMHO, it is human 'error' in the programming that makes each audio engine 'sound' different, the different anti-aliasing filters written for each, and maybe a little 'grass is greener' syndrome.

Maybe an expert DSP mathmatician will chime in. But it's really an age-old debate that you can find LONG discussions about on pretty much any audio forum online. ProSoundWeb is one place to go spelunking for info.

As far as moving to Nuendo, the real impedement to your workflow is going to be learning a new workflow!

[croyal]

Originally posted by Shooshie:

In fact, if you look at DP's chart, I think that it very clearly reinforces what my ears tell me when I record with 48k sample rate, as opposed to 44.1. The difference between 44.1k and 48k, for me, is night/day. 48k opens up the sound spatially and makes the audio clear and beautiful. 44.1k suffers from a "stuffiness" in DP which I've heard many people complain about. Well, that could very well be due to the problems inherent in their anti-aliasing filters beginning their cut-off slope at about 18KHz, as opposed to 22K.

Also, it looks to me that it would be better to use Barba-Batch or Spark to drop down to 44.1k for mastering to CD, as opposed to using any DAW for that purpose.

Shooshie

I can't hear any difference between 44.1k and 48k in my setup. It may be in some setups that a particular convertor is biased for one or the other and is doing a realtime SRC when playing at its other rate(s). On the Rosetta class convertors (old and new) they are 24 bit encoders that dither to 16 bits as needed- but they always sample initially at 24 bits. I know that bit depth is not the issue, but Sample Rate conversion may be an inherent part of some units/softtware. I suspect that some convertors simply sound better at one rate or another- taking into account all the variables- including the analog path for each ADC and DAC. For example, many years ago I used to complain about Cubase's sound when mixing- but mixing through an original 2408- so how could I really tell? Could softaware be biased to work better at particular rates as well?

As far as software anti-aliasing filters go, I believe it's more of a problem when SRC downwards. High speed and realtime internal software/hardware have to put that damned filter on at the end of the process. I used to use Alesis' AI-1 real time convertor from 48k to 44.1 This hardware piece made signals sound "rounded off" when converting to and from ADAT's at 48k.. Yet, the ADAT sound wasn't as compromised when material was recorded at 44.1 in first place- hence no need for the SRC.

As I'm doing more and more of these conversions, I'm begining to believe that most SRC software ultimately works like truncation of bits: to go from 24 to 16 bits you just whack off the last 8 bits and you're done. From 96k to 44.1 internally you change the rate, do a little math, slap on a filter to cover the mistakes and you're done. A more pleasing sound from dithering (noise) masks the bit truncation. And in the same way the inaudible noise added when doing convertor-to-convertor SRC retains high frequencies- without any anti-aliasing filter cutting off those exact same frequencies. If you mix OTB you have to resample anyway, and the CtC process also eliminates the need for dithering if just changing sample rates.

BarbraBatch is good, but to my ears not as good as a realtime transfer from one great convertor at 96k to another great convertor at 44.1. In audio, realtime processes seem to almost always sound better, whether it's tape duping or burning CDs or SRC. And, it usually sounds better to use separate machines- such as in tape-to-tape or convertor-to-convertor transfers. Tape duping machiines and (in the same way) SRC software are very useful; but nonetheless compromise the quality of the final product. The comparison page simply showed who did the least damage.

Chris

[Shooshie]

This little episode set me off to read again, and I found this page which tends to support what I was saying above. I was just guessing--"reasoning" I suppose, but with a lot of unsupported leaps--but this guy's explanations, though basic, seem to support my leaps if you go down to the section on "when do we need to dither?"

Shooshie

<small>[ August 17, 2005, 02:14 AM: Message edited by: Shooshie ]</small>

[Aramis]

Shooshie !!!!
My secret page

So !!!!!!
Aramis

[Timeline]

My songs were purely 24 bit 96K files. The OMF transfers did not embed audio or include plugs so rule that out. Just fader levels and mutes. When Nuendo opened the OMF it simply played back the SDII original files.

Unless you make a change to a file, all audio stays as original in the original DP folder. To prove that you can look in the audio folder Nuendo sets up as a NEW project and see there is nothing there.

If the target application's converting in real time & uses ASIO from MOTU for the IO's involved, there might be a difference in dither methods and I think it has something to do with that process. This is where It began to sound poor to me.

I did this on several mixes. I tried to EQ as well and it sounded bad as if truncated and filtered on the top end so I gave up and stayed with DP and finished the album.

Again, I'm not saying Nuendo's audio engine is to blame and I have not tried recording in N3 and comparing so it's likely that N3 does sound great when beginning and ending a project from the start there.

I think DP should be the first company to take into consideration the target applications internal floating bit methods and have an applications 'type specific' selection function for compatibility i.e. OMF ProTools, OMF Nuendo, OMF whatever....

[kwiz]

Originally posted by Aramis:
Shooshie !!!!

And just to add to the problem , I remember not hearing the same on different nights because I was more tired .....

So !!!!!!
Aramis

Ain't that the truth.