Amiga audio early lead lost..

[ChaosLord]

It simply doesn't mix 4 channels together. It mixes channels together in pairs to two different outputs. I.E. sure, you could set it up to output four channels without using any CPU, but nope, it doesn't mix them all together.

 True that it mixes 2 channels to the left and 2 to the right.  But it does not require you to use the CPU to resample all the sounds to a specific frequency.  It supports a giant range of frequencies and just mixes the different samples of different frequencies together like magic.  And it does the mixing on its own with DMA.  One DMA channel per DAC.

[Karlos]

Paula does fixed frequency of 22030hz in PAL or any of a wide range of other values.

None of which are correctly tuned for musical notes. Given that your argument was that paula is "playing" the music by using variable sample rate replay, this is pretty important. Humans with good hearing can typically resolve different pitches to within a few Hz. As you need shorter periods for higher frequency note play back, the error in 1/x for integer x becomes more apparent.

That said, Paula's imperfect reproduction (in many ways - quantisation noise, aliasing etc) is what makes it unique and I really do love it as an instrument in it's own right. However, for *production*, that is, the end result of combining individual instrument performances into a finished recording, all of those cool thing suddenly become a handicap.

And btw: I tried to convince Gunnar that Natami Paula needs a special fix to be able to generate 44100Hz exactly.  But he repeatedly rejected the idea as unnecessary and silly and pointless.  What is your opinion on that?

It's fine if you don't mind having, for instance, to replay video at the wrong rate to synchronize the audio or resample the latter to avoid issues.

[Karlos]

True that it mixes 2 channels to the left and 2 to the right.  But it does not require you to use the CPU to resample all the sounds to a specific frequency.  It supports a giant range of frequencies and just mixes the different samples of different frequencies together like magic.  And it does the mixing on its own with DMA.  One DMA channel per DAC.

It's not really mixing (at least in any data processing perspective). The analogue outputs are simply summed.

I would have much preferred if there were some software controllable left/right volumes (or just volume + pan) for each of the 4 channels. That would have been awesome.

[Linde]

True that it mixes 2 channels to the left and 2 to the right.  But it does not require you to use the CPU to resample all the sounds to a specific frequency.  It supports a giant range of frequencies and just mixes the different samples of different frequencies together like magic.  And it does the mixing on its own with DMA.  One DMA channel per DAC.

I'd say that analog summing is nothing like "magic" but I understand how someone with no idea about the subject might use that word.

Hardware mixing of multiple buffers with arbitrary pitches is a pretty basic feature of PC sound chips since more than a decade back, and before that, hardware wavetable synthesis with RAM big enough to easily accommodate the max total sample size of mods were prevalent.

DSPs, allowing CPU-free realtime effects like reverb, 3D positioning, dynamic compression, gain control, seamless and invisible resampling and mixing of a great number of input streams are ubiquitous. Whatever obscure feature the Paula has that a modern PC sound chip is missing was left out because it isn't necessary. Quite oppositely, they were there on the Paula because they made practical sense in a setup with very limited CPU and RAM resources. Frankly, I love the Amiga sound, but I'm not going to pretend that the Paula is technically superior to modern PC sound chips or even an AWE32 in any way. Horribly non-linear DACs, its totally unmatched architecture and its limitations is what gives it its great character, but it can't match your average on-board cheap PC sound card in terms of fidelity, features and flexibility. Even a lot of the features I've described that apply to most sound cards are frequently left unused because doing everything in software has so little overhead on a basic PC setup that it doesn't matter.

You're really awful at discussing these things, too. You fail to acknowledge when you are proven wrong and keep coming back to points that have already been adressed and refuted in previous posts. In terms of a metaphor, you very quickly paint yourself into a corner, but apparently you don't mind stepping in paint.

[minator]

Sound in CD players started at 44KHz for a good reason, playing samples at a lower frequency adds noise.

Paula works the same way as early digital instruments like the Fairlight or PPG.  However, as soon as the technology was cheap enough everyone switched to resampling because the quality was so much better.

Paula is an interesting 80's sound chip but the things that make it good are mainly due to the vast amounts of sound artefacts it produces.  This mean it's great if you want a coloured crunchy sound but it's pretty much useless for even normal audio playback these days.  Even the most basic PCs have been better for a very long time.

[AmigaClassicRule]

Sound in CD players started at 44KHz for a good reason, playing samples at a lower frequency adds noise.

Paula works the same way as early digital instruments like the Fairlight or PPG.  However, as soon as the technology was cheap enough everyone switched to resampling because the quality was so much better.

Paula is an interesting 80's sound chip but the things that make it good are mainly due to the vast amounts of sound artefacts it produces.  This mean it's great if you want a coloured crunchy sound but it's pretty much useless for even normal audio playback these days.  Even the most basic PCs have been better for a very long time.

I guess that concludes the discussion about Amiga audio. This is what I call a complete summary and conclusion in a topic.

[psxphill]

Sound in CD players started at 44KHz for a good reason, playing samples at a lower frequency adds noise.

Well sort of, it's mostly due to taking the highest frequency that people can hear then doubling it because of nyquist. The specific frequency was chosen because they wanted to used analogue video tapes (http://en.wikipedia.org/wiki/PCM_adaptor) to store the digital data on & that was the rate they ended up with (much the same as how they came up with the Amiga's maximum dma sample rate). It also allowed the 74 minutes they wanted in the size cd they wanted.
 
CD players have used filters, oversampling and 1 bit dacs to reduce the harmonics inherent in digital playback.
 
CD audio was over-hyped, it survived so long because people don't like format changes.

[Thorham]

Paula is an interesting 80's sound chip but the things that make it good are mainly due to the vast amounts of sound artefacts it produces. This mean it's great if you want a coloured crunchy sound but it's pretty much useless for even normal audio playback these days.

I find that when playing back CDs in 44Khz 14 bit stereo with calibration, there's not much coloring at all. In fact, I think most of the coloring in the 8 bit 4 channel mods is because of the used samples and tracker effects, and not because of the DACs.

Playing back CDs sounds perfectly fine, just not in 8 bit (the horror!) and you do need double scan modes for the 44 Khz playback.

[Karlos]

I find that when playing back CDs in 44Khz 14 bit stereo with calibration, there's not much coloring at all.

Well, the whole point of CyberSound's calibration (the driver for which is what Paula 14-bit mode in AHI uses) is to mitigate the non-linearity of Paula's DACs, so if you have calibrated it well, it should sound relatively clean. It does depend on what you listen to, however. I can still hear quantization noise and aliasing however, particularly on gentle volume ramps in quiet parts of music, even having painstakingly calibrated my machine. These days, it's hidden behind digital noise. I suspect some caps have dried out wherever the analogue stage is decoupled.

It doesn't bother me particularly however, since any multichannel mods I write end up rendered to disk and when I do use Paula as a sample player, I usually want all that colouring.

[paul1981]

Well, the whole point of CyberSound's calibration (the driver for which is what Paula 14-bit mode in AHI uses) is to mitigate the non-linearity of Paula's DACs, so if you have calibrated it well, it should sound relatively clean. It does depend on what you listen to, however. I can still hear quantization noise and aliasing however, particularly on gentle volume ramps in quiet parts of music, even having painstakingly calibrated my machine. These days, it's hidden behind digital noise. I suspect some caps have dried out wherever the analogue stage is decoupled.

It doesn't bother me particularly however, since any multichannel mods I write end up rendered to disk and when I do use Paula as a sample player, I usually want all that colouring.

Personally, I find even the uncalibrated 14 bit output sounds fantastic. Maybe I'm biased.

[Karlos]

Personally, I find even the uncalibrated 14 bit output sounds fantastic. Maybe I'm biased.

Again, it all depends what you use it for. Really, the intent of the 14-bit mode was to overcome the fidelity issues imposed by the existing 8-bit mode. Viewed in that light, you really ought to calibrate your CyberSound driver and use that calibrated output to get the best fidelity from your system.

[Thorham]

I can still hear quantization noise and aliasing however, particularly on gentle volume ramps in quiet parts of music, even having painstakingly calibrated my machine.

Yeah, there's still some noise in those soft spots.

These days, it's hidden behind digital noise. I suspect some caps have dried out wherever the analogue stage is decoupled.

Time for a re-cap perhaps?

It doesn't bother me particularly however, since any multichannel mods I write end up rendered to disk and when I do use Paula as a sample player, I usually want all that colouring.

But what is that coloring? Apart from some noisiness I don't hear any.

Personally, I find even the uncalibrated 14 bit output sounds fantastic. Maybe I'm biased.

Perhaps, but calibration should still reduce the noisiness of the soft parts of music.

Viewed in that light, you really ought to calibrate your CyberSound driver and use that calibrated output to get the best fidelity from your system.

It's a pity the calibration software doesn't allow calibrating the left and right channels separately, instead, all channels are calibrated at the same time

[psxphill]

But what is that coloring? Apart from some noisiness I don't hear any.

http://en.wikipedia.org/wiki/Digital-to-analog_converter
 
"The fact that DACs output a sequence of piecewise constant values (known as zero-order hold in sample data textbooks) or rectangular pulses causes multiple harmonics above the Nyquist frequency. Usually, these are removed with a low pass filter acting as a reconstruction filter in applications that require it."
 
The Amiga's low pass filter is far too low to leave enabled when you're outputting 44k. Even with normal rates it sounds too muddy. In fact back in the day I used one of these http://www.amiga-hardware.com/showhardware.cgi?HARDID=1447 which really boosted the high frequencies. I might open it to see what it contains, I think it might just be a single channel graphic equaliser.
 
A lot of people don't notice the crunchiness, whether this is due to lack of sensitivity to high frequencies (I can hear frequencies higher than is predicted for my age) or lack of interest I don't know.
 
I still have my ghettoblaster too (picture isn't mine... but it looks the same) and the fake surround and 3d bass really made the amiga sound great.
 

[B00tDisk]

Funny; just the other day I watched the LGR Oddware review of the Covox Speech Thing on Youtube...

I was struck by how much the sound output reminded me of the Amiga's audio.

Of course, I will now stand patiently and wait for the screaming to start...

http://www.youtube.com/watch?v=spOenlrSSOE

[Thorham]

The Amiga's low pass filter is far too low to leave enabled when you're outputting 44k.

That filter is useless.

A lot of people don't notice the crunchiness, whether this is due to lack of sensitivity to high frequencies (I can hear frequencies higher than is predicted for my age) or lack of interest I don't know.

Are you talking about 44khz 14 bit calibrated? If so, there is certainly some noise in the softer parts of music, which is MUCH worse without calibration. If not, then what do you mean with crunchiness?

I was struck by how much the sound output reminded me of the Amiga's audio.

That's just because of the samples they used. That typical Amiga sound is caused by the samples used in modules.

[ChaosLord]

The Amiga's low pass filter is far too low to leave enabled when you're outputting 44k. Even with normal rates it sounds too muddy.

 I agree.  I always turned my filter off and left it off.

However at some point a new Amiga came out that had much brighter audio.  It sounded sooooo much better.  I can't remember if it was the A3000 or A1200 but it was one of those.  I donno what they did but the audio was noticeably brighter.

I still have my ghettoblaster too (picture isn't mine... but it looks the same) and the fake surround and 3d bass really made the amiga sound great.

I donno what 3D bass is but I have always used a Sony stereo with a "Surround Sound" button that makes the Amiga's audio sound fantastically awesome!  It completely removes that wide separation thing and makes listening on headphones a wonderful audiophonical experience.