Amiga "14-bit audio" refuted

[daxb]

Originally Posted by no9

Recommended for what? I wouldn't recommend using DigiBoosterPro in 8 bit mode to anybody if there is a hypothetical 14-bit available for Paula. It was quite common opinion amongst Amiga musicians that those '14-bit HIFI' AHI modes not necessarily sound so close to 16-bit as raw mathematical calculation would suggest, but it was still improvement over 8 bit. And Thomas measurements seems to confirm that notion.

Most music is/was done in 8 bit with Protracker/Octamed and some other trackers and that was/is the only choice for games. Hence the word recommend.

DigiboosterPro is another story (except Paula DMA mode the channels are mixed using AHI). If you use a sound card then use the 16 bit AHI sound driver. If you use Paula then use a 14 bit stereo++ driver. With a slow CPU (<040) use the 14 bit FAST modes (using pre-calc tables) to be able to mix enough channels. With 040/40 (my setup) I recommend the normal 14 bit stereo++ mode. The 040/40 is too slow for the HIFI modes (4 channels is already too much). I don't know but guess that a 060 is also too slow for 6-8 channels or more. I would say nobody used the HIFI modes. However, the normal 14 bit mode sounds good but is far away from 16 bit.

There are 12-bit samplers. And there is software where you can reduce sound parameters according to your wish.

I know but I've never seen 12 bit samples and nearly nobody used it. So, it is a very special environment. However, I asked for 14 bit. Somebody knows how 11 bit or 13 bit sound?!

[ross]

Originally Posted by chb

There's a very comprehensive document by Henryk Richter on Amiga audio and also on calibration (section 4.2)

I have just finished reading the paper: it is done very well!
It is a sum of many things scattered in various threads also here on EAB, and it is a great reference on how Paula really works, refuting many misinterpretation that are often spread..

[no9]

Originally Posted by daxb

Most music is/was done in 8 bit with Protracker/Octamed and some other trackers and that was/is the only choice for games. Hence the word recommend.

Oh, recommended by 'democracy' not by what quality demands, right? So what this has to do with this thread at all? Btw. most Amiga game music and sfx was made with low quality sounds like 11kHz and even lower. Is this also kind of recommendation what to use now?

DigiboosterPro is another story (...) I don't know but guess that a 060 is also too slow for 6-8 channels or more.

Well... as in the past I've made couple of tracks with DBpro I dare to vaguely remember that there was plenty of options to choose from like mono/stereo modes, mixing frequency and this resulted with the variable number of channels available there for given CPU. Also usage of so called 'DSP echo' effects etc. So I'm not sure where are you going with all these recommendations and what it has to do with the topic.

This is the thread about output of theoretical 14-bit sound quality on Paula, and definitely not about what most or nobody used.

[daxb]

The topic (discussed several times) is about 14 bit on Amiga through Paula is (real) 14 bit or less and different opinions to this. Thomas did some measure. Most judge by opinion. I doubt that most users know how 14 bit or 12 bit or 10 bit sound. Where is the reference to compare to? For me it is irrelevant if 14 bit Paula playback is 14, 12 or 10 bit. It is somewhere between 8 and 16 bit. You can argue about it but in the end it is all the same.

[Oxygene]

Early CD players (like those based on TDA1540 series DACs) were 14 bit. Could be used for testing and comparing, I suppose.

[ross]

Perhaps first it is better to decide which machine to try Paula with.

Value for the fixed LP RC filter (from posted paper):

- Amiga 500/2000: 4.42 kHz (100 nF, 360 ohm)
- A600 Rev2: 27.7 kHz (3.9 nF, 1.5k ohm)
- A1200 Rev1d: 27.7 kHz (3.9 nF, 1.5k ohm)
- A1200 Rev2: 34.4 kHz (6.8 nF, 680 ohm)
- A4000: 4.52 kHz (47 nF, 750 ohm)

The A2000 doesn't seem the most suitable machine to me...

[Thomas Richter]

For the test frequency used here (441Hz), we are well away from the critical frequency of the filter of any of the machines. Since an ideal R/C filter is linear, the output of a sine wave (such as generated) is a sine wave again, just with a different amplitude, and with a phase shift. Both are compensated by the small test program. Whatever filter is in the audio output, it does not matter as long as it is a linear filter.

[ross]

Originally Posted by Thomas Richter

For the test frequency used here (441Hz), we are well away from the critical frequency of the filter of any of the machines. Since an ideal R/C filter is linear, the output of a sine wave (such as generated) is a sine wave again, just with a different amplitude, and with a phase shift. Both are compensated by the small test program. Whatever filter is in the audio output, it does not matter as long as it is a linear filter.

Now I don't want to go into endless arguments, but the frequency used is really unfair for Paula. I understand your reasons but being an arbitrary value you could also choose it differently. But here I stop myself.

[no9]

Originally Posted by daxb

The topic (discussed several times) is about 14 bit on Amiga through Paula is (real) 14 bit or less and different opinions to this. Thomas did some measure. Most judge by opinion. I doubt that most users know how 14 bit or 12 bit or 10 bit sound. Where is the reference to compare to? For me it is irrelevant if 14 bit Paula playback is 14, 12 or 10 bit. It is somewhere between 8 and 16 bit. You can argue about it but in the end it is all the same.

At least no recommendations this time. I posted link to a tool that let's you play with bits. Not much mystery there. But if you need to know more then here I edited sounds with decreased bits depths from 16 to 1. Second set has slight white noise in the background because it makes difference.

[ Show youtube player ]


And here more bits and what it is all about
[ Show youtube player ]

[ross]

A not too complex article on bit-depth and sound quality:
https://www.soundguys.com/audio-bit-...plained-23706/
and why for a casual listener 12 bit audio may even be good.

[nikosidis]

All the tech. talk might be interesting for some, but the proof is in what you are hearing.
I played guitar since I was a kid and I have a high-end stereo sound system. Music and sound been a interest of mine forever. Paula sound natural with CD quality material as long as your screen mode is over 30Hz.
22khz Audio with PAL high-res screen mode or similar is not good enough for my taste. I'm talking real voice, instruments. For modules 8 bit, lower frequency is no problem. Paula is for sure no contender for a high-end hi-fi system but for casual audio playback most would be happy. I also tried many Amiga audio players and the best is EaglePlayer. There are quite big differences in the playback software. Old Amiga hardware like the A500 I have does not sound good. My Amiga 600 and 1200 sound much better.

[bloodline]

Originally Posted by ross

A not too complex article on bit-depth and sound quality:
https://www.soundguys.com/audio-bit-...plained-23706/
and why for a casual listener 12 bit audio may even be good.

Yeah, 12bit is just enough to keep the noise below normal hearing, and certainly well below any music mix.

I still have a Roland W-30 here which uses 12bit sampling and it’s difficult to tell it isn’t 16bit, unless you do a specific A/B test.

[Estrayk]

I promise that without encouraging any controversy, I made a video some time ago comparing my Atari Falcon with real sound at 50Khz 16bits against my A600 (14bits 50Khz). Just in case someone is interested in appreciating the difference by listening to what Thomas has discovered. Maybe someone can find it interesting......

[ Show youtube player ]

[pandy71]

Originally Posted by Thomas Richter

I don't care what people claim. I'm a physicist. I care what I can measure.

So follow industry standards...

[Thomas Richter]

Originally Posted by ross

Now I don't want to go into endless arguments, but the frequency used is really unfair for Paula. I understand your reasons but being an arbitrary value you could also choose it differently. But here I stop myself.

Then chose it differently. The tools I posted allow you to specify the frequencies. Though the phase drift aka incorrect frequency is not the real problem here. It's compensated.

[Thomas Richter]

Originally Posted by pandy71

So follow industry standards...

I'm not even sure what you want to say here.

[bloodline]

Originally Posted by Estrayk

I promise that without encouraging any controversy, I made a video some time ago comparing my Atari Falcon with real sound at 50Khz 16bits against my A600 (14bits 50Khz). Just in case someone is interested in appreciating the difference by listening to what Thomas has discovered. Maybe someone can find it interesting......

[ Show youtube player ]

I remember watching this a while back, it’s a nice comparison (though I’m not sure the choice of reference track was particularly helpful), also I think the mpeg compression is clearly evident (but YouTube is YouTube...). Anyway the difference between the two machines is extremely clear when I play back using my Adam A7X nearfields.

[ross]

Originally Posted by Thomas Richter

Then chose it differently. The tools I posted allow you to specify the frequencies. Though the phase drift aka incorrect frequency is not the real problem here. It's compensated.

Problem is that I've no more a real machine..

But I clearly remember the abysmal difference on audio quality between my old A500 and my A1200 (1d4), with the same conditions of play of course.
Especially with regard to high-frequency reproduction (for modules with low frequency samples in many cases I preferred the A500, but this by the different R/C stage causing more aliasing artifact on A1200).

So it's not only a question of filter, but also of better/newer components, better D/A separation or less noise or who know..

I would do a test on an A1200

[roondar]

Originally Posted by bloodline

I remember watching this a while back, it’s a nice comparison (though I’m not sure the choice of reference track was particularly helpful), also I think the mpeg compression is clearly evident (but YouTube is YouTube...). Anyway the difference between the two machines is extremely clear when I play back using my Adam A7X nearfields.

This is a very nice comparison. It shows (or sounds ) that there is indeed a clear audible difference between the two, I can hear it even on my rather old Logitech speakers and it's very, very clear on headphones. But to me, it also shows that the Amiga's "14 bit" audio output quality is clearly 'good enough' for the system - it sounds really quite impressive if you keep in mind this is a 1985 audio chip that was never designed to do more than 8 bit samples to begin with.

Somewhat strangely, to me the difference seems far more clear in the frequency response and not so much in the areas I'd expect. I'd expect to hear a lot more noise on the Amiga output, instead I mostly heard a difference where the Falcon was hitting the "high" tones "better".

[chb]

Originally Posted by roondar

Somewhat strangely, to me the difference seems far more clear in the frequency response and not so much in the areas I'd expect. I'd expect to hear a lot more noise on the Amiga output, instead I mostly heard a difference where the Falcon was hitting the "high" tones "better".

Same for me, I mainly hear a different frequency response. The A600 sounds quite muffled.
While in theory the A600 and the A1200 should have the same fixed filter, it seems like that's not always the case in reality, but the filter is equivalent to that of the Amiga 500 with 4 kHz cutoff:
http://eab.abime.net/showthread.php?t=86880
Maybe that also applies here?