Who else was enchanted with their A1200?

[roondar]

I liked my A500 when I had it and used it lots.

But I loved my A1200 after I got it. Got it towered up with a Fast ATA expansion, CD-ROM drive and a Blizzard 1230 Mk IV. Sure, the framerate it got in Alien Breed 3D II wasn't good, but I still played that game to death on it - quite atmospheric. Great stuff, still use it today - though nowadays I have changed the specs a bit - different IDE expander, SATA->PATA converted DVD drive, Indivision AGA Mk3 and a Rapidroad USB. And of course, it's been recapped

Still my most loved Amiga, use it weekly or more.

Ironically, I like coding for the OCS systems more.

[Thorham]

Originally Posted by Karlos

You can hear the "ringing" very clearly in the sine bass.

Indeed.

Originally Posted by Karlos

In this example it's probably unwanted and if I had to guess, it's an example of the sinewave having a relatively short period (say 32 or 64 samples) being played way down the bottom octaves, so a pretty extreme example and a good justification for per channel filters

Yes, but how about an example where it's wanted? Also, why have your equipment color the sound if you can color the samples? Seems more practical.

[Karlos]

Originally Posted by Thorham

Indeed.
Yes, but how about an example where it's wanted? Also, why have your equipment color the sound if you can color the samples? Seems more practical.

It's not that was intended to colour the sound the way it does, it just does. It's the "natural character" of the instrument. When you are making music and you have analogue signal paths, everything colours the sound in some way or another.

Notice that I use the word instrument here. When I was making music, my Amiga was a sequencer, but also an instrument. As an instrument, I may have it routing through outboard processors that do more to the sound.

When you are using it as a direct tracker only then you do have choices to make. Do you want the samples sounds to be more faithful reproduction of the source? If so, then you need to make sure you capture them, ideally with a good quality 16 bit sample and carefully downsample to 8 bit at whatever frequency range you can accept. This may mean using multiple samples over octaves or even over smaller note intervals. You are now treating Paula as a basic rompler. Or, if you are wanting a more harsh electronic sound, maybe you do the exact opposite and lean into lofi with all the characteristics it has. Personally, if I was using a sinewave bass, I'd generally have a long period loop because I don't want those ringing alias overtones for that. But if I'm using a saw type sound, it's quite nice.

I probably have some music somewhere that makes use of this, but it isn't going to sound like basic tracker music because I'm using the Amiga as an instrument, with external processing and ultimately multitracked.

[minator]

Originally Posted by Thorham

Indeed.
Yes, but how about an example where it's wanted?

Here's an emulated example, it's a track I did a while back where I was trying to get this effect.
I used TAL Sampler and messed with the DAC settings to emulate an especially grungy Paula-esq sound.


ATOM 13

[Karlos]

A

Originally Posted by minator

Here's an emulated example, it's a track I did a while back where I was trying to get this effect.
I used TAL Sampler and messed with the DAC settings to emulate an especially grungy Paula-esq sound.


ATOM 13

You absolutely nailed it in the percussion and baseline. Half of the overtones on that bass sound sound like the aliasing effects I'm describing. The quantisation noise on the fade of some of the samples is a bit OTT, but I assume that's intentional.

[Karlos]

Originally Posted by Karlos

I did but not digital. Lost on tape somewhere.

Unless... I'll have to review some old CDR / HD

So, this isn't the configuration described, but using Paula as a rompler.

The music for one of my MC64K videos features a number of high quality multisampled instruments (acoustic guitar notes, bass and percussion) and a field recording of some bird sounds from a dawn chorus. The side guitar parts are each recorded using 4 Paula channels. Chord voicings on the fake strum involve a 5th chord sample to increase the apparent polyphony. The bass and melody part were recorded together duophonically (bass part on two channels to the left, melody to the right). The percussion I think was two 4 channel tracks and the bird sounds were a final layer.

All layers were recorded separately with realtime noise reduction, reverb etc. courtesy of using outboard effects processing. The layers were captured and mixed in Sound Forge with no real adjustments except some level tweaking.

[ Show youtube player ]

By contrast this other video usesa 12 channel MMD3 (Octamed SS) file. Some samples are ripped from a demo. Maybe more than one. Edit: not direct to disk, I mean not post produced. It was played directly through the MU100 AD inputs for a touch of EQ and maybe reverb, the result was capture direct to disk and done. Replay was 14-bit stereo at the closest match to 48kHz it allowed.

[ Show youtube player ]

What I've not been able to locate yet is a digital capture of any of the Paula as a mono synth with the external filters. That era, I don't think I had any kind of disk recorder. By the time I had that, I was using digital outboard effects, including filters.

[aeberbach]

Tell you what, the Pistorm32-lite has completely remade the Amiga 1200 for me. I'm 6 levels into Eye of the Beholder again. I can't believe how fast everything is! RTG graphics are amazing, though I am not seeing much of that yet while in games.

[Thorham]

Okay that's cleared up then. Guess I'm so used to using Amiga audio the way it works on the peecee that I never really noticed anything interesting about it.

[Karlos]

Originally Posted by Thorham

Okay that's cleared up then. Guess I'm so used to using Amiga audio the way it works on the peecee that I never really noticed anything interesting about it.

I think it's a common enough sentiment. The reason for those two video choices is to highlight it. The second track would have benefitted from a soundcard because we aren't using Paula to play individual sounds, we are using it to replay a 16-bit steam as well as it reasonably can. OctaMED doesn't use calibrated 14-bit replay so even above the 27kHz limits, there's all kinds of artefacts and noise that is not present in the multi track recorded first example. Even so, it's still passable.

A lot of what people think of as the Amiga sound is due to 4 channel mod music. A lot of the charm comes from the limitations of 4 hard panned channels and the discipline necessary to arrange music that sounds good when you are confronted with them and things like note chopping. If you want to do, say a guitar solo with a delay effect on it, your only choice normally is to replay the same notes on another channel slightly later and at a lower volume. Suddenly you are down to 3 effective channels. If your machine only had 512K, you might be low on memory for samples too, so you're using lower fidelity samples than you might otherwise, increasing all the noise and aliasing. Even playing chords,you use a sample - maybe you even used the chord generator feature in your tracker. This reduces the fidelity even further because in a basic triad chord each note can't be louder than 1/3 the available dynamic range to avoid clipping. Naively (and this isn't quite correct) you can say that each individual note in the chord is only 8/3 bits resolution overall, but more importantly the noise tends to be additive (if you've used a chord generator and not sampled the chord).

Things are completely different when Paula is playing one part, using all 4 channels, with high quality samples, all mixed into a single mono track that you intend to mix afterwards, as I hope the first example shows. Paula can hold her own against much more expensive samplers and romplers in that context, despite the 8-bit sample depth.

[grond]

Originally Posted by Karlos

Even playing chords,you use a sample - maybe you even used the chord generator feature in your tracker. This reduces the fidelity even further because in a basic triad chord each note can't be louder than 1/3 the available dynamic range to avoid clipping. Naively (and this isn't quite correct) you can say that each individual note in the chord is only 8/3 bits resolution overall

I think this should be ld(256/3) which would be something like 6.4 bits of resolution. But I get your point.

[Karlos]

Originally Posted by grond

I think this should be ld(256/3) which would be something like 6.4 bits of resolution. But I get your point.

Erm yeah. That's what I meant. The dynamic range per note in a triad chord sample is only 1/3 the full range possible for a single note. No matter how carefully you construct the chord, it's not as good as three independent voices at the full dynamic range, especially if the individual note volumes are different for expression purposes.

This is another place where Paula kicks other 8-bit solutions where it hurts. Although the sample data are only 8-bit, the full 8-bit range is preserved across the entire channel volume range. Your basic single PCM channel 8-bit sound blaster can't touch it. You need at least a 16 bit output DAC for software mixing to be equivalent or better fidelity.

[Cobe]

I was!