Author Topic: Autoselection of instruments and other questions from a newcomer  (Read 332 times)

mptntguru

  • Guest
Re: Autoselection of instruments and other questions from a newcomer
« Reply #15 on: November 03, 2017, 20:53:46 »
xmms gives louder high frequences than polyphase but polyphase is more musical and pleasant to ear

Offline Saga Musix

  • OpenMPT Developers
  • *****
  • Posts: 6,292
  • aka Jojo
    • Download music, samples, VST plugins: Saga Musix Website
  • Operating System: Windows 7 x64
Re: Autoselection of instruments and other questions from a newcomer
« Reply #16 on: November 03, 2017, 22:18:04 »
The answer is a bit vague for me. Do you mean that in order to hear for example drums (fx2) instead of piano (fx1) while playing the midi keyboard i have each time to switch to an instument tab and select number 2 at the top, and to hear piano to go back to instrument tab again and select digit 1? Or do such switching in the "plugin" section of patterns tab at the top which seems easier. Still i don't understand why you can't implement automatic choosing the right instrument assigned to the given channel when i select a cell in that channel. It would save efforts definitely.
You do not need to go to the instrument tab, you can directly change the active instrument on the pattern tab e.g. from drop the instrument dropdown list at the top or through keyboard shortcuts. There are also shortcuts to pick up the last used instrument on a channel (find "Pick up nearest instrument number" in the keyboard configuration).
Admittedly that's not the best workflow when working with MIDI, but it's also on the roadmap to have better (automatic) MIDI recording capabilities in the future. It's a lot of work to implement though, so don't expect that to happen very soon.

Plugin delay compensation is also a lot of work to implement. You have to be aware that OpenMPT is the spare-time project of two developers, and we do not earn any money from this. All of these features you mention are useful without doubt, but they are also very complex to implement, and while many of them have been considered before, there is often a reason why they have not been implemented yet (availability of spare time being one of many reasons).
» No support, bug reports, feature requests via private messages - they will not be answered. Use the forums and the issue tracker so that everyone can benefit from your post.

mptntguru

  • Guest
Re: Autoselection of instruments and other questions from a newcomer
« Reply #17 on: November 04, 2017, 00:26:10 »
May be you can try to get that patreon sponsorship which sponsors airwindows. Your program beats all others I have tried as to sound quality and i get the loudest volume without distortion.  Of course it's first impressions and i need to try mixing more tracks for full comparison but I swear that kontakt sounds 100% unprocessed like raw samples. And also it's the first software that i have found to have 32bit integer processing. There are forum topics where superiority of integer over float is explained. One guy told me that  float creates kind of antialiasing effect which smashes details. May be you can also incorporate 64bit integer to beat 48integer of protools. The winner was saw studio which had 64bit integer and was considered best sounding daw. Unfortunately it's 32bit and has no possibilities for rewire.
You can implement choosing FX number of an instrument by a combination like holding winkey and pressing numbers on numerical keyboard.

I have read an article that digital technology handls high frequencies badly vs analog. That's really so, even if i raise highs both on a track and on master I still get don't get enough highs and low mids dominate with the entire mix muddy. If you cut down lows you lose energy and get flat space. The only acceptable equalizer which boosts highs enough is free dtblkfx though it generates some distorting frequencies it's the only acceptable.
Also an interesting observation that waves rverb from 2001 sounds like crap on modern daws but fantastically in openmpt. That's real magic... I suspect that later daws have kind of hidden protection against quality from old plugins. They either reject them during scanning or produce untolerable result. Openmpt has most neutral coloration, sequoia has protruding warm color and too soft sound. I was quite satisfied with DP but it started to reboot my pc at 1\4 cpu load. Sequoia users have to use external hardware mixers to get some decent sound. All other daws I have tried are just not worth mentioning. Windows drivers sound much better than linux. That's why I have moved from linux back to windows after 4 hard years of fighting for quality on linux. One guy told that you need to use fft-based plugins for effects to get studio quality. This seems to be the truth since fft kind of reproduces infinity. So far I use only dtblkfx and need to find some more of the same type. As for reverb the best one is VB aphro and for delay it's h3000. May be this info will be of help to those seeking studio quality at home. I would stay away from plugins younger than 2001 unless there is urgent need.

Offline manx

  • OpenMPT Developers
  • *****
  • Posts: 158
Re: Autoselection of instruments and other questions from a newcomer
« Reply #18 on: November 04, 2017, 10:54:59 »
Your program beats all others I have tried as to sound quality and i get the loudest volume without distortion.  Of course it's first impressions and i need to try mixing more tracks for full comparison but I swear that kontakt sounds 100% unprocessed like raw samples.

No, OpenMPT does not magically produce better sound with a given VST than any other audio production software. If I left this claim uncommented here, I (and we, as developers of OpenMPT) would implicitly insult the developers of any other audio software of doing incredibly stupid things to the audio signal. We will not do that. They do not do stupid things. And, unless you can actually prove your claims, you should immediately stop publishing them because you are also just insulting other developers by your very claims.

And also it's the first software that i have found to have 32bit integer processing. There are forum topics where superiority of integer over float is explained. One guy told me that  float creates kind of antialiasing effect which smashes details. May be you can also incorporate 64bit integer to beat 48integer of protools. The winner was saw studio which had 64bit integer and was considered best sounding daw. Unfortunately it's 32bit and has no possibilities for rewire.

Unless you provide the reasoning why integer would supposedly be better than floating point, I cannot really comment on that. It just makes no sense whatsoever, every single detail in that whole paragraph.
64bit integer processing is just a waste of resources, unless you have about 2 billion channels you are mixing simultaneously (at which point quantization noise could actually matter here ... but you have seriously other more important problems at that point).
Also, to get the facts straight, OpenMPT uses 32bit integer processing for its mixing, of which there are 28 bit fractional precision and 4 bit clipping headroom. VST Plugins always use 32bit floating point, because that's the way VST works. This will internally get converted to the aforementioned integer format. Always.

I have read an article that digital technology handls high frequencies badly vs analog. That's really so, even if i raise highs both on a track and on master I still get don't get enough highs and low mids dominate with the entire mix muddy. If you cut down lows you lose energy and get flat space. The only acceptable equalizer which boosts highs enough is free dtblkfx though it generates some distorting frequencies it's the only acceptable.

If you EQ a track and can hear the change on the single track/channel, you will have exactly the same change (to that track) in the mixed master signal.
If your EQ cannot properly amplify the highs in the signal, it is a bug or deficiency in the particular EQ you are using. If the EQ generates additional harmonic distortion, it is also not working properly. (I am not claiming that any particular EQ does or does not have such problems).
The applies equally well to analog as it does to digital audio.

Also an interesting observation that waves rverb from 2001 sounds like crap on modern daws but fantastically in openmpt. That's real magic... I suspect that later daws have kind of hidden protection against quality from old plugins. They either reject them during scanning or produce untolerable result.
DMO plugins like Waves Reverb can be used in either 16bit integer or 32bit float mode (maybe even others, I did not investigate that). It might be the case that the implementations of that particular effect differ in quality (Saga Musix may be able to comment on that). OpenMPT 1.26.01 started running DMO plugins in 32bit float mode rather than 16bit integer like it did before (see https://openmpt.org/openmpt-1-26-01-00-released ). It might be the case that other DAWs use different processing formats. You can easily test that by applying the exactly identical plugin effect settings to the exactly identical input signal and render the result to a wave file and compare. The result will never be bit-exact identical because of different internal mixing representation of the audio signal and possibly dithering, however the difference will be in the quantization noise at maximum. Again, there is nothing fundamentally magic happening here.
Also, if a given DAW runs DMO plugins in 16bit integer mode, you can easily introduce clipping distortion if you apply reverb to an already close to 0dBFs input signal. Such clipping distortion is of course easily audible. The fix is also easy: Attenuate your signal properly before the effect in order to avoid clipping in the effect.

Windows drivers sound much better than linux.

That's just wrong.

One guy told that you need to use fft-based plugins for effects to get studio quality. This seems to be the truth since fft kind of reproduces infinity.

This also makes no sense. Whether a particular effect is implemented via a FFT is merely an implementation detail. It certainly has implications about what the effect can do to the audio signal, but that's all there is to it. There is no inherent advantage in applying each and every effect via FFT, in fact, it would be incredibly difficult and close to impossible for some effects to even try doing that.

So far I use only dtblkfx and need to find some more of the same type. As for reverb the best one is VB aphro and for delay it's h3000. May be this info will be of help to those seeking studio quality at home.

There are good and there are not so good plugins out there, like with any other software. I cannot comment on the particular ones you mentioned because I do not know them.

I would stay away from plugins younger than 2001 unless there is urgent need.

Why? That makes no sense whatsoever at all.

mptntguru

  • Guest
Re: Autoselection of instruments and other questions from a newcomer
« Reply #19 on: November 04, 2017, 17:51:47 »
[qoute]VST Plugins always use 32bit floating point, because that's the way VST works.[/quote]
I have read that waves corp makes 48bit vsts for protools or may be those are rtas or au format but they are claimed to be 48bit. Also I have got a comment on another forum that even if your daw is 64-bit, vsts use 32-bit internal processing as you have correctly noticed but vsts with 64-bit processing also exist though they are very rare.

Quote
They do not do stupid things
Daws have different latencies and stability. For example with sequoia\samplitude i get much lower latencies than with other programs. It allows to have small latencies with large buffers because they somehow use additional software buffers besides those set in audio driver. Where other program choke sequioa still goes on. Another program claimed to have that positive thing is studioone. I remember reading on a linux forum that large latencties with large buffers is a kind of commercial limitation. i would trust to my hearing first of all.
Quote
No, OpenMPT does not magically produce better sound with a given VST than any other audio production software.
Then please comment on the description on schism website that schism provides studio quality without expensive hardware. Taking into account that openmpt has similar algorithms why can't i  conclude that it provides studio quality as well? And also why do some sound designers prefer to use openmpt where they still have an option to use top commercial software?

Concerning superiority of integer over float http://www.sonicstudio.com/pdf/papers/48FixedVs32Floats.pdf
Though even 24-fixed point is better than 32-float which is actually 23-bit (mantissa). "The first observation is that digital filtering when we allow the user to select high-Q, very low-frequency filters is difficult at the best of times. Even 64-bit floating point can produce significant error energy if the best filter forms are not used. Even for floating point, it is important to use forms that have normalized state variables so that imbalances in the state values do not lead to further degradation of the precision of the result. Clearly, the performance of 32-bit floating point and 24-bit integer will be considerably inferior to that of 64-bit floating point, so we might conclude that it is not possible to achieve high-quality results for these extreme filter settings. Furthermore it is shown that sweeping the settings of a filter with time excites some aberrant behavior when the state variables are not normalized, even with 64-bit floating-point arithmetic. 48-bit integer is proposed as a compromise between economic realizability and ultimate precision. The increased headroom and guard bits allowed by the format provide enough precision to allow some extreme filter settings and still preserve a 24-bit result after several stages of processing."

We have been persuaded that floating-point is a kind of progress which is not so which is the reason why I choose 24bit integer instead of 32float in daws nowadays.
So you can introduce 48 integer into openmpt for more professional sound but still 32 integer beats 32 float alltogether.  Anyway vst technology is not trustworthy for me because it does not allow integer processing. By the way can directx plugins use integer processing? If yes than it explains why dx plugins from 1999-2001 may sound better than vsts. The waves rverb which i mentioned is DX converted to vst via dxshell program.  i am almost sure that i should stick to DX or use dxs wrapped to vst by dxshell. I am also curious if standalone kontakt uses integer processing because it can be part of explanation why it sounds better than in vst form.

Quote
You can easily test that by applying the exactly identical plugin effect settings to the exactly identical input signal and render the result to a wave file and compare. The result will never be bit-exact identical because of different internal mixing representation of the audio signal and possibly dithering
Due to different processing schemes daws give different hue to sound which i call "lack or presence of studio like sound or studio like quality". And the only daw which sounds strict and serious like studio is DP. Another one was emagic logic for windows before apple bought it. One guru told me that you should use only hofa and airwindows vst plugins to preserve that quality but after talking to another guru I have learnt other secrets and am switching to dx 1999-2001 + openmpt.

mptntguru

  • Guest
Re: Autoselection of instruments and other questions from a newcomer
« Reply #20 on: November 05, 2017, 00:11:37 »
11. Is there a master channel and can i put a plugin to it?

Offline Saga Musix

  • OpenMPT Developers
  • *****
  • Posts: 6,292
  • aka Jojo
    • Download music, samples, VST plugins: Saga Musix Website
  • Operating System: Windows 7 x64
Re: Autoselection of instruments and other questions from a newcomer
« Reply #21 on: November 05, 2017, 00:22:40 »
Yes, you can check the "master" checkbox in the plugin settings. Note that master plugins must be in higher plugin slots (e.g. FX100 and upwards) than other plugins or otherwise the plugin chain will possibly not work as intended.

I am not going to comment on any of the esoteric "program X sounds better than program Y  because it uses technology Z" stuff. You prove to have a lot of dangerous half knowledge and do not want to accept to be corrected on those points anyway.
» No support, bug reports, feature requests via private messages - they will not be answered. Use the forums and the issue tracker so that everyone can benefit from your post.

Offline manx

  • OpenMPT Developers
  • *****
  • Posts: 158
Re: Autoselection of instruments and other questions from a newcomer
« Reply #22 on: November 05, 2017, 10:05:40 »
Quote
VST Plugins always use 32bit floating point, because that's the way VST works.
I have read that waves corp makes 48bit vsts for protools or may be those are rtas or au format but they are claimed to be 48bit. Also I have got a comment on another forum that even if your daw is 64-bit, vsts use 32-bit internal processing as you have correctly noticed but vsts with 64-bit processing also exist though they are very rare.

VST3 (which is not supported by OpenMPT in any kind) also supports 64bit floating point. I was talking about VST2 though, which is the most widely used form of VST plugins, and VST2 uses mostly 32bit floating point. VST2 gained 64bit floating point support in version 2.4. Version 2.3 and earlier do not support 64bit floating point. OpenMPT uses only the 32bit variant.
Edit: corrected VST version numbers.


Daws have different latencies and stability. For example with sequoia\samplitude i get much lower latencies than with other programs. It allows to have small latencies with large buffers because they somehow use additional software buffers besides those set in audio driver. Where other program choke sequioa still goes on. Another program claimed to have that positive thing is studioone. I remember reading on a linux forum that large latencties with large buffers is a kind of commercial limitation. i would trust to my hearing first of all.

Now you are talking about latency, which has nothing to do with audio mixing quality. Yes, different DAWs and audio APIs can achieve different levels of low latency. Latency is only relevant for live realtime playback and completely orthogonal to mixing quality, which can be observed by just rendering to a file and involves no realtime latency whatsoever at all.


Quote
No, OpenMPT does not magically produce better sound with a given VST than any other audio production software.
Then please comment on the description on schism website that schism provides studio quality without expensive hardware. Taking into account that openmpt has similar algorithms why can't i  conclude that it provides studio quality as well? And also why do some sound designers prefer to use openmpt where they still have an option to use top commercial software?

You probably can conclude that, if you want. Even though I would be hesitant to call the OpenMPT (and Schism) resampler truely studio quality because of the amount of aliasing it introduces when downsampling. It is good enough for almost all practical purposes though.

What you absolutely cannot conclude is that other software is not also "studio quality" just because they did not put that fancy term on their website. It's not a precisely defined term, and thus rather useless for quality assessment anyway.


Concerning superiority of integer over float http://www.sonicstudio.com/pdf/papers/48FixedVs32Floats.pdf

This paper for the most part talks about implementation of filter coefficients in particular. It is not a given to use the same precision or datatype for filter coefficients as is used for the audio samples itself. Not directly relevant or related.


Though even 24-fixed point is better than 32-float which is actually 23-bit (mantissa).

IEEE 754 binary 32bit floating point has 24bit effective fractional precision.

And no. Neither one is generally better than the other. It totally depends on the actual use case (and in audio mixing, there are use cases for both).


We have been persuaded that floating-point is a kind of progress which is not so which is the reason why I choose 24bit integer instead of 32float in daws nowadays.

Almost no software will allow you to actually switch the internal mixing datatype. For almost all cases you are merely selecting the final output format (either for writing to file or for talking to the soundcard driver).


So you can introduce 48 integer into openmpt for more professional sound but still 32 integer beats 32 float alltogether.

No. There is no point to it.


Anyway vst technology is not trustworthy for me because it does not allow integer processing. By the way can directx plugins use integer processing? If yes than it explains why dx plugins from 1999-2001 may sound better than vsts. The waves rverb which i mentioned is DX converted to vst via dxshell program.  i am almost sure that i should stick to DX or use dxs wrapped to vst by dxshell.

VST technology is totally fine and perfectly adequate for professional music production. Heck, any professional recording uses VST nowadays. If it would sound as bad as you imply, dont you think people would have complained?

Also, OpenMPT has native DMO support. There is absolutely no point in using any DX/DMO to VST wrapper.


Quote
You can easily test that by applying the exactly identical plugin effect settings to the exactly identical input signal and render the result to a wave file and compare. The result will never be bit-exact identical because of different internal mixing representation of the audio signal and possibly dithering
Due to different processing schemes daws give different hue to sound which i call "lack or presence of studio like sound or studio like quality". And the only daw which sounds strict and serious like studio is DP. Another one was emagic logic for windows before apple bought it. One guru told me that you should use only hofa and airwindows vst plugins to preserve that quality but after talking to another guru I have learnt other secrets and am switching to dx 1999-2001 + openmpt.

This is just esoteric bullshit. Sorry for being harsh, but I have described precisely how to measure it and you continue to make dubious claims without any proof. Go measure it.

DX/DMO plugins from 1999-2001 are not magically better than current VST plugins. As already explained, for historic reasons there is a great chance that DMOs are run in 16bit integer mode, which will easily introduce clipping or bad signal-to-noise ratio. Also, audio algorithms have greatly improved over the last 15 years. Plugins from ~2000 cannot have that knowledge. Additionally, computer processing capabilities have also increased manyfold, and thus current plugins can do more complex and accurate calculation than older ones (not saying that they all do, but older ones certainly cannot because they did not have the required processing power available to them).


I will not comment any further on this topic. In fact, I will have to lock this topic in order to stop you from continuing posting wrong, bogus and false claims which do confuse novice users. You are doing active harm here.


You are of course free and welcome to ask further questions about how to use OpenMPT though. We will happily answer them. However, I suggest starting a new topic for each question (i.e. a forum topic per topic - it's called that way for a reason) as that will make discussions easier to follow.
« Last Edit: November 05, 2017, 11:28:56 by manx »