Home   News   Concept   AMIGA-Compatible   Hardware   Forum   Questions+Answers   Pictures   Contact & Team

Welcome to the Natami / Amiga Forum

This forum is for AMIGA fans interested in the new NATAMI platform.
Please read the forum usage manual.



All TopicsNewsQAFeaturesTalkTEAMLogin to post    Create account
Do you have ideas and feature wishes? Post them here and discuss your ideas.

Sound: (Not So?) Next Generationpage  1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 
Claudio Wieland
Germany
(Natami Team)
Posts 703
16 Mar 2012 09:31


IIRC, USB has 1ms time slots. Hence a latency of maximum 1ms, which is not bad at all. It's stupid drivers on stupidly slow-poke OS's that collect a couple frames and only then proceed. Should not be a problem on Amiga type computers, though ;p .

Rafael Rafafredd
Brazil

Posts 21
19 Mar 2012 15:29


I may be wrong, but I think that whenever the audio processing takes place on the main CPU, and not on dedicated CPUs, buffers are required for dealinbg with the OS priority when executing all with the same CPU, like audio, OS functions and graphics. So, some latency is always introduced by this buffer. In a professional situation, this latency got to be perfectly measured, reported to the driver and compensated for.
 
And no, it is not a Windows only "feature".

When you are dealing with trackers, where audio only comes out, it is never a problem. Problems comes into place when you use it for professional recording, where audio should be able to go in, be processed in realtime, and out, with the least perceived latency possible. It means 5ms or lower (10ms net I/O latency maximum).

Marcel Verdaasdonk
Netherlands

Posts 3974
19 Mar 2012 16:25


Rafael That is WHY i mentioned the 3Dcode since you would not be doing both 3D operations on the screen and do your audio operation at the same time!

In a game this is different because this is nine out of ten times pre-determent of what happens when and thus does not need to happen real time.

Rafael Rafafredd
Brazil

Posts 21
20 Mar 2012 01:24


I see. But if nearly realtime (less than 1ms latency) would really work, then audio signal DSP processing would have to be processed by 3Dcore only, and never by main 68k processor. Is that what you suggest?

Rafael Rafafredd
Brazil

Posts 21
20 Mar 2012 03:45


By the way, if someone wants to write drivers professional audio interfaces for Amiga OS, contact TC Electronics and Focusrite. I know both have cooperated with FFADO LINUX driver sending the team all the info and official docs, while many other companies won't even reply to a request like this.

Marcel Verdaasdonk
Netherlands

Posts 3974
20 Mar 2012 04:14


Rafael rafafredd wrote:

I see. But if nearly realtime (less than 1ms latency) would really work, then audio signal DSP processing would have to be processed by 3Dcore only, and never by main 68k processor. Is that what you suggest?

That is exactly what i mean.

Besides Reuse of powerful units is a good thing right when they are not in use, right?

Golgoth 27
France

Posts 185
20 Mar 2012 15:09


Rafael rafafredd wrote:

By the way, if someone wants to write drivers professional audio interfaces for Amiga OS, contact TC Electronics and Focusrite. I know both have cooperated with FFADO LINUX driver sending the team all the info and official docs, while many other companies won't even reply to a request like this.

Thanks for the advice, it's great to see so much people focused on the same project. It's when people are united like the NatAmi team, Aros, and some other, that the projects can become ambitious.

Guillaume Michalakakos
France
(MX-Board Owner)
Posts 454
20 Mar 2012 16:34


I totally agree with Joe M : It is IMHO far better to focus on Natami's sound chip Pauline before talking about external solutions.
 
  I think we have to ask ourself first what could/will be the use of the NatAmi in audio. Let me explain in details how I see this as a musician and sound enginer :
 
  - I have trouble to see the NatAmi become the center of a professional audio workstation even if, and I believe we all know this already, this is technically possible. The first and biggest problem will be the support for the professional external audio hardware. We can of course forget protools HD hardware, and I'm almost 100% sure that there will not be a support from MOTU, RME, Metric Halo, Universal Audio, Euphonix, Apogee, ... except miracle. Another problem will be the compatibility with some plug-ins standart, and after the compatibility, the CPU power needed to run those plug-ins :-D. Another problem will be with the Hard Drive : If you want to record f.e. 8 channels in 24bit-192khz at the same time, I'm not sure that even the fastest IDE drive will follow. Perhaps with a RAID... As you can see it will be hard for the NatAmi here. But I see some areas where the NatAmi could be really interesting and powerfull (please note that I will test at least two of those sollutions) :
 
  - For me the most important : Home-Studio recording ! First I see the Amiga as the perfect personnal computer (this is just my opinion), and I think we should focus on personnal uses first. NatAmi already have good quality audio-in connectors. If the A-D converter is good, and I think he will, then you already have a good stereo-in channel (or two mono channels). Most of the time in a home-studio configuration you will not record more than one audio channel at the same time since your house is not a professional recording place.
  What you need to add to this is a good ANALOG mixer. Let me explain what are the advantages in my opinion :
  you don't need drivers,
  analog mixers are easier and more intuitive to use than a numeric sound card and software configuration.
  For a good audio quality, you will have to pay less money for an analog sollution. Good quality in ANALOG audio start in low prices (my MACKIE 402-VLZ3 is a very good exemple that cost about 110 EUR only)
 
  I will tell you a story : some time ago a friend was asking me to come/see his personal recording (voice + electric guitar) in order to know if he was making this correctly. I came with my personal very cheap nomad recording sollution that I wanted to test : MACKIE 402-VLZ3, and an apple powerbook pismo G3-400mhz, with a mini-jack audio-in, 16bit 44,1khz max. (and of course some microphones, that wasn't used here). My friend had a powerfull core2duo PC with a M-AUDIO external sound card (don't remeber the model exactly, but know the price was around 350-400 EUR). The A-D conversion was 24bit-96khz. All what my friend did was good, I had nothing to change. But after using his sollution and mine using the SAME microphone at the same place, we heard that my sollution was better : more dynamic range (yes, in 16bit vs 24bits !!!!!!!), less background noises, efficient EQ BEFORE the digital conversion... So less work to do after the record and no need for an external DI about the electric guitar...
  As we can see, there is a way to have a nice and cheap home-studio solution with the NatAmi who don't need any external driver or support.
  Of course there is certainly good numeric sound card that are "open" and that can be used for more channels in, but we still need drivers. So I think like Joe that we should investigate this sollution after.
 
  - It is THE add-on for the Home-Studio recording : MIDI instruments. As I said there is no problem to have very low latencie using MIDI instrument and let the computer generate the sounds. Well, amiga OS already support MIDI as well and has a lot of possibilities here. Of course there is still the old MIDI boxes in serial port, but I think what we need is support for modern midi boxes : that means drivers for USB MIDI boxes. But perhaps posseidon already support some ?
 
  - True computer music using trackers. This is of course an area where NatAmi will perform good. I'm not an expert in trackers but I agree that ADAT in-out could be a nice sollution.
 
 
 
Joe M wrote:

    When will we be able to download some hi quality audio samples recorded from a Natami developer board? Guillaume, have you done some testing yet?
 

 
  For now audio-in don't work on the NatAmi yet since there is much more important to achieve first. So I didn't made any record test now. I already wrote a music piece (electric bass, electric & accoustic guitar, violin, viola, double bass (not sure now, but I hope the musician will be available) / synth and drum in midi) that I want to record in a pure home-studio way (understand : recording channels one by one, at home) by connecting my little 402-VLZ3 mixer to the NatAmi audio in. After that I was thinking to add the midi instruments to go more further in home-studio style : midi instruments will be connected to NatAmi who will generate the sounds.

Rafael rafafredd wrote:

By the way, if someone wants to write drivers professional audio interfaces for Amiga OS, contact TC Electronics and Focusrite. I know both have cooperated with FFADO LINUX driver sending the team all the info and official docs, while many other companies won't even reply to a request like this.

 
This is indeed very interesting. Thank you ! 
 
 

  No expectations about NatAmi and DMX ? ;-)

Marcel Verdaasdonk
Netherlands

Posts 3974
20 Mar 2012 17:40


Guillaume I think i posted my opinion of DMX512 in another thread here but i suppose it won't hurt repeating it a little.

We don't need to imitate a feature that has no function in a creative machine.
I see the Natami as a tool for people to use their artistic ability in whatever form it maybe.

It's not just about old games!

So what does this mean in Audio.
That we need to be able to generate and record Audio in a good Resolution.
For me this means more then CD quality since well CD's have a aliasing problem with it's 16 bits, 44.1KHz.
24 bits, alone doesn't solve this!

Sadly i don't see any ail in using a CIA time for this since a sample rate close to 196KHz one needs to divide the CIA's clock by about 3.63 rounded up.

To be honest i am not all that into recording i am more into the reproduction of Audio hence love for Vinyl albums.

Besides that assuming we hijack the system for audio recording theoretic max of IDE/ATA is 133MB/s same with PCI thus we will have some overhead.
Hm, 8 Channels 24Bits 196KHz sampling, let's say 8 channels, times 32bits, times frequency, divided by 1048576 for MB/s.
I end up with about 50MB/s so it ought to be within reach but the question remains is it feasible?(did i forget something?)

Sorry all for blurting this core dump on here just my 2 cents for what it's worth.

SID Hervé
France

Posts 663
20 Mar 2012 18:40


Hello

I agree with this orientation which is more realistic. I think Home computer = Home Studio (export will always be possible).

OFF TOPIC :

Personally, I always preferred the track by track (named here as the home studio). This offers more flexibility. But yes, I know, it is more tedious.

Thanks


Megol .

Posts 671
21 Mar 2012 14:43


Marcel Verdaasdonk wrote:

Guillaume I think i posted my opinion of DMX512 in another thread here but i suppose it won't hurt repeating it a little.
 
  We don't need to imitate a feature that has no function in a creative machine.
  I see the Natami as a tool for people to use their artistic ability in whatever form it maybe.
 
  It's not just about old games!
 
  So what does this mean in Audio.
  That we need to be able to generate and record Audio in a good Resolution.
  For me this means more then CD quality since well CD's have a aliasing problem with it's 16 bits, 44.1KHz.
  24 bits, alone doesn't solve this!
 
  Sadly i don't see any ail in using a CIA time for this since a sample rate close to 196KHz one needs to divide the CIA's clock by about 3.63 rounded up.
 
  To be honest i am not all that into recording i am more into the reproduction of Audio hence love for Vinyl albums.
 
  Besides that assuming we hijack the system for audio recording theoretic max of IDE/ATA is 133MB/s same with PCI thus we will have some overhead.
  Hm, 8 Channels 24Bits 196KHz sampling, let's say 8 channels, times 32bits, times frequency, divided by 1048576 for MB/s.
  I end up with about 50MB/s so it ought to be within reach but the question remains is it feasible?(did i forget something?)
 
  Sorry all for blurting this core dump on here just my 2 cents for what it's worth.

Unless you are super-human 16bit, 48KHz for output should be enough. AFAIK there are some 1M USD waiting for so called audiophiles* that can prove those kind of super-human abilities. Higher precision is useful for internal processing to avoid audioable precision losses

(* I'd call myself an audiophile but not of that kind)

Megol .

Posts 671
21 Mar 2012 14:49


Claudio Wieland wrote:

IIRC, USB has 1ms time slots. Hence a latency of maximum 1ms, which is not bad at all. It's stupid drivers on stupidly slow-poke OS's that collect a couple frames and only then proceed. Should not be a problem on Amiga type computers, though ;p .

Simplified yes. Low speed USB HID have 100Hz update frequency IIRC.
A good USB stack should have no problem with latency nor processor load.

Marcel Verdaasdonk
Netherlands

Posts 3974
21 Mar 2012 18:28


Megol
Aliasing right?
  The more Samples the less risk you have picking the wrong frequency.
 
  At 48KHz you have in the upper range a human ear pick up about 3 samples, on a single period this thus can be interpreted as a different frequency.(overtone, aliasing)
  This for me is a problem, hence that i prefer a higher sample rate.
  Besides that the reason i picked the highest usable sample rate is because one needs to take the worst case when looking at a problem.
  This can be used in math and answer if something is acceptable.
 
  IMHO being able to record 8 tracks simultaneous at 196KHz 24 bit resolution is quite a feat for a home computer system.
 
  To be honest for professionals the Resolution ends somewhere between 20, 21 bits, Besides that I rounded up to 32 bit for a reason.(Higher then that it is more work to clean up Noise)
  Besides that you can never over sample on a master, hence the high sample frequency.(it's easier to reduce samples then it is to add afterwards)

SID Hervé
France

Posts 663
21 Mar 2012 19:12


Megol . wrote:

...Higher precision is useful for internal processing to avoid audioable precision losses

This is true. Moreover, the set of frequencies to be reproduced is always variable. So to avoid the use of costly "What can do more can do less" the approach proposed by Guillaume Michalakakos is reasonable. Different factors must be considered, and to name one: the probable impacts on the overall project.

Nicholas Blachford

Posts 1
21 Mar 2012 23:10


Guillaume Michalakakos wrote:

Another problem will be with the Hard Drive : If you want to record f.e. 8 channels in 24bit-192khz at the same time, I'm not sure that even the fastest IDE drive will follow.

Dude, what century did you buy your hard drive in?
192/24 X 8 is 4MB /second.  My *old* video camera did that to *tape* 10 years ago!

The question is, will it run ProTools, Cubase, Logic, Reason, Reaper etc?
If not I'd forget about being the centre of a studio large or small.

PCs and Macs and even iPads can do high sample rates and high bit rates these days, there's no point competing with this.  You wont offer anything everyone else can't do in better quality, faster and cheaper.

You can run a modern studio with hundreds of channels of music, synths, samplers, drum machines and effects galore all on a single laptop.  Don't believe me?  Listen to the charts, a lot of big artists do exactly that.

If you want to get noticed by musicians be different.

The stuff becoming popular these days is lo-fi. Musicians happily pay a lot of many to make things sound bad.

The Amiga had a truly unique sound chip but for the most part completely failed to use it.

People think Paula is a 4 channel sample player but it can do a lot more.
They don't realise you can chain the channels to do FM synthesis, or AM, or both.
Paula also generates notes by altering the sample playback rate - this gives the Amiga a really distinctive sound, especially with lower rate samples.

An 8 or 16 channel version of Paula would be interesting, bizarrely, 8 bit 28KHz will be just fine.
Another one is to do the FM and AM in better quality / bitrate.  Who is making FM sound chips these days?

Combine that with a tracker with MIDI in and you've got something that is truly unique.  I might even buy one of those.


Jakob Eriksson
Sweden
(Moderator)
Posts 1097
21 Mar 2012 23:29


Ehm, yes Nicholas, but we are talking sustained to IDE in an Amiga clone. Maybe it will be easy, I don't know, but the CPU has to do all that and talk to the card. The Natami is very fast, in some things actually faster than a PC, but not CPU fast.

Nixus Minimax
Germany

Posts 272
22 Mar 2012 09:18


Marcel Verdaasdonk wrote:
At 48KHz you have in the upper range a human ear pick up about 3 samples, on a single period this thus can be interpreted as a different frequency.

No, it can't. Your theory makes me cringe.



Marcel Verdaasdonk
Netherlands

Posts 3974
22 Mar 2012 09:44


Nixus 15KHz is within human hearing range it has a bit over 3 samples at a 48KHz sample rate.

I hope i explained it better in this way.

What i mean is that this can cause trouble while aliasing.

Nixus Minimax
Germany

Posts 272
22 Mar 2012 10:44


Marcel Verdaasdonk wrote:

Nixus 15KHz is within human hearing range it has a bit over 3 samples at a 48KHz sample rate.
 
  I hope i explained it better in this way.
 
  What i mean is that this can cause trouble while aliasing.

OK, perhaps I just don't understand what you are saying. In this case I wish to apologise for my rude comment. 15 kHz is in the human hearing range. But what do you mean with "it has a bit over 3 samples at a 48 kHz sample rate"? Are we talking recording or replay? In the case of recording I agree that a sampling frequency of 48 kHz will mean a very high demand for analogue filtering. For this reason 192 kHz recording is popular as the problem of analogue input filtering is relaxed a lot. After digitalisation you can put a very high order digital low-pass filter on the sampled data and down-sample without problems. For the same reason replay at 48 kHz is absolutely sufficient.


Marcel Verdaasdonk
Netherlands

Posts 3974
22 Mar 2012 11:10


I wish there as a ability to post pictures here it's easier to draw out for me.

Basically i have been speaking about Aliasing here.
What i mean is that between these points which were sampled you can draw the overtone of that frequency without the points altering position.
This is what Aliasing basically is you cannot see the difference between frequency and it's overtone if your sample rate is too low.

This is something one could apply to both Recording and reproduction of a wave form.(But Recording is more important in this matter)

I wasn't even talking about filtering.

To be honest I still think 28KHz has it's charm. ;)


posts 304page  1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16