Hertz Overload - new demo for AGA Amigas

[saimo]

Originally Posted by copse

As a noob it makes me wonder how do all the other music playing routines on the Amiga deal with this? Is it not relevant for them for some reason? Or is this something that becomes obvious to the ear when dealing with your higher resolution data?

I have no idea of what other solutions do
The synchronization is relevant only when chasing high quality, so I'd be surprised if for "normal" use cases like modules playback the tracker routines care about this issue at all.

[saimo]

Originally Posted by Karlos

That doesn't seem likely. Surely, when the volume is set to zero, there shouldn't be any sound to hear at all?

Yes. My explanation sucked, so I'll try again (but I have to rush it as I'm again in a break at work).
I tested the output of the very last, quiet, part of the coda of the demo music, repeated for 14 seconds on my A1200:
* volume 1: there is some barely audible noise that requires concentration and headphones to hear;
* volume 0: the noise is very audible.
This means that even if the period is 50 and even if the volume counters get reloaded when a new sample gets written to AUDxDAT (as the other test proved and also as logical), a volume of 1 still has some effect, whereas I'd have expected not to work at all, as the volume counter can never reach 0 in 50 clocks.
Could it be that the volume counter actually starts from 0 and then wraps around from 63?

I'm not sure if the concept is still clear, so just to recap, the original idea was that your source 16 bit audio (for example) would be preprocessed into a set of frames (minimum size of 2 samples) of 8 bit that have been scaled so that it can be played back at a particular volume to re-approximate what the 16 bit sample levels would be. This 2:1 sample to volume data rate was a side effect of how channel modulation works and the fact that Paula DMA fetches 16 bits. My original experiments were all in pure software and used a bigger ratio, e.g 16:1 as this reduces the stream size quite a bit and didn't degrade the output too much.Paraj (or another gentleman of quality) did implement the proposed HW playback mechanism in a prototype, which sounded pretty good to me, but unfortunately, that was all under emulation as my real HW is not currently in a working state.

Yes, the concept was clear. And the advantage when playing back with the CPU is that it's possibile to have a modulating volume for each sample (1:1).
Later I hope to whip up a proggie that plays back 8 bit waveforms and to make also a version that does the volume modulation you proposed.
By the way, I find it a really nice idea!

[Gorf]

Originally Posted by saimo

I see. I had not connected your question to the volume issue (because I had not yet thought about the volume!)
...
... maybe there's only so much that can be done with 8 bits? Make no mistake, I'm impressed that 8 bits can sound so good at 192 kHz, but that's too taxing for a poor stock A1200.

Yes - the limitations of the 14bit method with oll its technical details was discussed a t thread in the Coders section here on EAB:
It will not work as intended for higher feeding rates.

One possible solution that was AFAIK not tested yet is to split the sample into two 8bit ones (per side) depending on the sound frequency: one channel for mids and highs and one channel for bass.
The bass channel can run on lower feeding rates and use just regular audio DMA, while the mids and highs are fed to the other channel by CPU at much higher feeding rates.

So you would use all 4 hardware channels at full volume and avoid the mentioned problem and the CPU demand would be the same as 2 of the channels are just using regular DMA.

[saimo]

Originally Posted by Gorf

Yes - the limitations of the 14bit method with oll its technical details was discussed a t thread in the Coders section here on EAB: It will not work as intended for higher feeding rates.

There's no link, but anyway I see that these matters had been discussed to death in various threads already (I feel so stupidly naive!).

One possible solution that was AFAIK not tested yet is to split the sample into two 8bit ones (per side) depending on the sound frequency: one channel for mids and highs and one channel for bass.The bass channel can run on lower feeding rates and use just regular audio DMA, while the mids and highs are fed to the other channel by CPU at much higher feeding rates.So you would use all 4 hardware channels at full volume and avoid the mentioned problem and the CPU demand would be the same as 2 of the channels are just using regular DMA.

Ingenious and advantageous in several ways. But isn't it subject to noise when the signal is weak anyway? For some reason (it must be its "asymmetry"), I doesn't entice me, though.

[Gorf]

Originally Posted by saimo

There's no link, but anyway I see that these matters had been discussed to death in various threads already (I feel so stupidly naive!).
Ingenious and advantageous in several ways. But isn't it subject to noise when the signal is weak anyway? For some reason (it must be it's "asymmetry"), I doesn't entice me, though.

The bass channel would be more susceptible to 8bit quantization noise, as it is fed a the usual 28kHz ... but on the other hand this is feeding rate is way beyond the Nyquist freq for any typical bass (0-250Hz)

Sadly we can not apply a low-pass filter per channel on the amiga - otherwise we could work with noise shaping for the bass as well ... but maybe some dithering would still work.

So it would need some testing how 8bit quantization noise is perceived by the human ear for bass frequencies and if we would get audible artifacts from dithering...

[Gorf]

Originally Posted by saimo

There's no link

Sorry, had to find it again myself first:

https://eab.abime.net/showthread.php?t=105744

[paraj]

Originally Posted by saimo

Yes, the concept was clear. And the advantage when playing back with the CPU is that it's possibile to have a modulating volume for each sample (1:1).Later I hope to whip up a proggie that plays back 8 bit waveforms and to make also a version that does the volume modulation you proposed.By the way, I find it a really nice idea!

Found old quick&dirty test source for that. Assumes 16-bit LE samples.

[saimo]

@Gorf

Originally Posted by Gorf

Sorry, had to find it again myself first:
https://eab.abime.net/showthread.php?t=105744

Thanks, I'll have a look.


@all

I just finished preparing the tools. Now just some time to test them on the A1200 (I've been working in WinUAE), prepare the package and then some basic instructions.

[saimo]

Alright, here are the tools for you to fiddle with: https://www.retream.com/_temporary/AQA.lha

A little premise...
To make the playback as fast as possible, I designed three very basic, chunk-based, uncompressed file formats which I called "AQA" = "Amiga Quality Audio" (pun related to WAV intended):
* AQA0: stereo 8-bit
* AQA1: stereo 14-bit
* AQA2: stereo 8 bit carrier + 6 bit signal 2:1 AM
(Sorry, for the moment being there's no documentation.)

Now, some quick instructions...

The archive contains 6 programs to be run from shell:
* cWAV8toAQA0: converts an unsigned 8-bit WAV to an AQA0
* cWAV16toAQA1: converts a signed 16-bit WAV to an AQA1
* cWAV8toAQA2: converts an unsigned 8-bit carrier WAV and an unsigned 8-bit signal WAV to an AQA2
* sAQA0/1/2: streams an AQA0/1/2.

68020+ required.

The command line parameters are self-explaining.

cWAV8toAQA0 and cWAV16toAQA1:
* WAVFILEPATH=WAVFP/A
* AQAFILEPATH=AQAFP/A
* TEXTFILEPATH=TFP/K

cWAV8toAQA2:
* CARRIERWAVFILEPATH=CWAVFP/A
* SIGNALWAVFILEPATH=SWAVFP/A
* AQAFILEPATH=AQAFP/A
* TEXTFILEPATH=TFP/K

sAQA0:
* AQAFILEPATH=AQAFP/A
* PLAYBACKFREQUENCY=PF/K/N
* PLAYBACKPERIOD=PP/K/N
* FASTINTERRUPTHANDLER=FIH/S
* FILTERON=FO/S
* VOLUME=V/K/N
* QUIET=Q/S

sAQA1 and sAQA2:
* AQAFILEPATH=AQAFP/A
* PLAYBACKFREQUENCY=PF/K/N
* PLAYBACKPERIOD=PP/K/N
* FASTINTERRUPTHANDLER=FIH/S
* FILTERON=FO/S
* QUIET=Q/S

If no frequency or period is specified, the one indicated in the AQA file is used.
Fiddling with those parameters allows to see how far the machine can be pushed. At some point, playback will start to stutter. In the worst case, the machine will freeze (adding safety code impacts radically the performance, so I didn't include any). Anyway, to freeze the machine, really small periods are required: sAQA1, on my A1200 in stock configuration, hangs at period 30 (118229 Hz) and, when the Blizzard 1230 IV is on, it hangs at period 11 (322445 Hz).
Speaking of performance, the FASTINTERRUPTHANDLER option tells the player to bypass the system handler of the level 4 interrupt and use a really tight interrupt handler (by the way, the aforementioned figures come from this option). If not specified, a system-compliant handler is used, which performs much worse. For safety, the channels get allocated first through the audio.device in all cases.
During the playback nothing shows (as the video DMA channels get turned off), except crazy colorbars when the system interrupt handler is used.
Playback can be interrupted by pressing (and keeping pressed for a little while) the right mouse button.

Finally, here are some test files, which I derived from this one provided by Sony:
* 55420 Hz 16-bit WAV
* 96000 Hz 8-bit WAV
* 55420 Hz 8-bit WAV to be used as carrier
* 27710 Hz 8-bit WAV to be used as signal (it's just a 1 Hz sinusoid)
* 96000 Hz 8-bit AQA0 (obtained from 96000 Hz 8-bit WAV)
* 55420 Hz 14-bit AQA1 (obtained from the 55420 Hz 16-bit WAV)
* 55420 Hz 14-bit AQA2 (obtained from the carrier and signal WAVs)

(Given this has been written in the usual haste - I shouldn't be sitting here now - I'm pretty sure numerous edits will follow )

[pandy71]

Originally Posted by saimo

That would be great In turn, tomorrow I'll prepare a modifed version of the playback tool that simply plays 8 bit data, so that anybody can make tests.

I will also provide SoX scripts so anyone can try to do conversion for samples playable with your player.

Originally Posted by saimo

Edit: relatively to the dither + noise shaping tests, I forgot to mention that I listened to the resulting waveforms directly on the PC where the processing was done, not on the Amiga (so the results were not affected by the Amiga's hardware).

PC HW is limited - samples are usually resampled to one of "standard" sampling frequencies. How good is resampler and how some other things are made may directly impact audio quality - i strongly recommend use Amiga HW for final judgment.The most important thing in noise shaping is lowpass filtering to remove noise shaping noise - usually everything above 20kHz should be suppressed as much as possible to avoid ultrasonic and out of band distortions.As Amiga audio HW has low pass filter then it should be "given for free".

[Gorf]

Originally Posted by pandy71

The most important thing in noise shaping is lowpass filtering to remove noise shaping noise

There is still "dithering": add some noise at high resolution (16bit, 24bit) to reduce quantization noise at low resolution (8bit).

It's a very old concept - Paper from 1964:
https://ieeexplore.ieee.org/document/1088973

More practical explanation:
https://www.allaboutcircuits.com/tec...on-distortion/
(What is Dithering? Using Dithering to Eliminate Quantization Distortion)

[saimo]

Originally Posted by Gorf

Sorry, had to find it again myself first:

https://eab.abime.net/showthread.php?t=105744

Last night I half-read those 15 pages (which I had stumbled upon in the past, but did not follow as I was not into this stuff back then).
Main takeaways for me:
* channels phase shift: apart from the DMA slots issue (which is irrelevant here), was it eventually fully understood? IIRC somebody wrote that it's due to the DACs;
* volume counters counting upwards: this answers my question about how it was possible that a volume of 1 has some effect also with a period of 50; given that the AHRM uses "volctrrld" (or something like that - I don't have it at hand now), I had assumed they were counting downwards.

Also, now that I know how the volume counters work and that they get reloaded also in CPU-driven operation, I can see that in this case of 14-bit playback with period 50 the error that affects the second sample of each word containing the least significant 6 bits is that it gets sent to the DAC with a delay of 14 clocks, so for such an interval of time the upper byte and the lower one go out of sync.
That said, despite this error and the imprecision of the 14-bit method, the result (at least on my A1200) is very good to the ears (I've tried sampling the output by connecting the A1200 to this PC, but there's a remarkable noise floor even when no sound is output).

[saimo]

Originally Posted by pandy71

i strongly recommend use Amiga HW for final judgment.

OK, I will (hopefully later today, after updating the tools, as the AM ones have stupid 0-padding in the signal data, although the player can be modified to remove that padding - and thus save 25% of space - at no computational cost).

[pandy71]

Promised script to convert SoX supported formats (flac can be used so feel free to feed your 192kHz 24 bit precious recordings) - hope script is self descriptive (just kidding - i'm not software developer so probably it can be done in more efficient and elegant way but it works) - i prefer noise shaping curve called "e-weighted-improved" as it should provide smallest overall noise in audible part - this is outcome of sub-optimal (improper) noise shaping filter from Amiga perspective - original filter is designed for 48kHz sampling where to get best results Paula should use highest possible oversampling to reduce overall noise level (exchange of sampling rate vs dynamics) - i use frequency 64489Hz (period 55 decimal) for sample rate in Paula (can be changed - just put decimal period value).

With used noise shaping filter this is measured noise floor (source was silence at 192kHz 32 bit float converted with help of the provided script to 64489Hz 8 bit) - with Paula optimal filter this will be probably way better.




There is also two gentle (+6dB) highpass (high shelf) filter to compensate two things - first embedded Amiga low pass filter - as default set Amiga 500 4.42kHz and second named preemphasis (1/8 sampling rate) - to compensate overall low pass filter (reconstruction?) in Amiga - both together provide approx 12dB boost so using setting with filter ON is recommended - should be neutral in overall response.

I use also sharp bandpass filter from 16Hz to 32kHz (overkill probably - can be removed especially if your goal is to use extremely high oversampling rates - but this will be waste of bits to be honest especially if noise shaping filter will have suboptimal characteristic).

Due high level of ultrasonic signal (noise) i recommend to apply 20..22kHz low pass filter at the amplifier input or use moderate power level to not burn your tweeters (or use speakers insensitive to excessive high levels of ultrasonic signals)

Spectrograms are produced for source and for converted file - they are Amiga friendly i.e. using 16 colors and size for 640x512 screen, IFF SVX can be produced but it is poor choice for stereo (for example foobar 2k plays stereo 8svx at mono twice longer i.e. same as channel order is stored in file).

@setlocal

@REM Where to Find SoX
@set SoX=C:\sox
@set PATH=%PATH%;%SoX%

@set PAL=28375160
@set NTSC=28636363

@rem SET CLK=28375160

@noise shapinf filters - some of them are designed only for 44.1kHz, some also for 48kHz - as Paula use different freq then frequency warping will occure - filter characteristic will be shifted - transposed - depends on how far from primary 44.1k/48k sampling rate is located.

@rem lipshitz, f-weighted, modified-e-weighted, improved-e-weighted, gesemann, shibata, low-shibata, high-shibata
@set noiseshp=improved-e-weighted

@rem LowPassFilter F=
@rem Amiga 500/2000: 4.42 kHz (100 nF, 360 ?)
@rem A600: 4.42 kHz (100 nF, 360 ?)11
@rem A1200 Rev1d: 27.7 kHz (3.9 nF, 1.5 k?)
@rem A1200 Rev2: 34.4 kHz (6.8 nF, 680 ?)
@rem A4000: 4.52 kHz (47 nF, 750 ?)

@SET LPF=4420

@SET PERIOD=55

@SET /a CLK=%PAL%/8

@SET /a AFREQ_I = (((100*%CLK%)/(1*%PERIOD%))+1) / 100
@SET /a AFREQ_F = (((100*%CLK%)/(1*%PERIOD%))+1) %% 100
@SET AFREQ=%AFREQ_I%.%AFREQ_F%

@rem SET /a APREC_I = (((100*%CLK%)/(4*%PERIOD%))) / 100
@rem SET /a APREC_F = (((100*%CLK%)/(4*%PERIOD%))) %% 100
@rem SET APREC=%APREC_I%.%APREC_F%

@SET /a APREC_I = (((100*%CLK%)/(8*%PERIOD%))) / 100
@SET /a APREC_F = (((100*%CLK%)/(8*%PERIOD%))) %% 100
@SET APREC=%APREC_I%.%APREC_F%


@SET file=%1
@SET fname=%~n1

@ECHO file=%file%
@ECHO filename=%fname%
@ECHO:
@ECHO --------------------------
@ECHO:
@ECHO Samplerate=%AFREQ%Hz
@ECHO:
@ECHO --------------------------
@ECHO:
@ECHO Precompensation=%APREC%Hz
@ECHO:
@ECHO --------------------------
@ECHO:
@ECHO Amiga LowPass freq=%LPF%Hz
@ECHO:
@ECHO --------------------------
@ECHO:

@pause

@rem SET aproc= 
@rem SET aproc=sinc 16-13.85k treble +6.0206 %LPF% .5 treble +6.0206 %APREC% .5 compand 0.25,1 6:-inf,-90.1,-inf,-90,-75,-70,-70 -3.0103 -90 0.20

@SET aproc=sinc 16-32k treble +6.0206 %LPF% .5 treble +6.0206 %APREC% .5 gain -n -0

@rem setting max signal level with headroom for noiseshaping
@rem set clip=
@rem set clip=gain -n -1.769537
@rem set clip=gain -n -2.110204
@rem set clip=gain -n -2.153742
@rem set clip=gain -n -2.197500

@set clip=gain -n -0.526579

@rem mono
@rem sox --multi-threaded --buffer 524288 --input-buffer 524288 -S -V4 -D -G %file% -e float -b 32 -t f32 "%fname%.f32" remix - rate -v -s -I %AFREQ% %aproc% stats -b 8 stat
@rem sox --multi-threaded --buffer 524288 --input-buffer 524288 -S -V6 -D -c 1 -r 48299 -e float -b 32 -t f32 "%fname%.f32" -c 1 -b 8 -e signed-integer -t s8 "%fname%.s8" %clip% dither -f %noiseshp% -p 8 stats -b 8 stat 
@rem sox --multi-threaded --buffer 524288 --input-buffer 524288 -S -V6 -D -c 1 -r %AFREQ% -b 8 -e signed-integer -t s8 "%fname%.s8" "%fname%.8svx"
@rem sox --multi-threaded --buffer 524288 --input-buffer 524288 -S -V6 -D -c 1 -r %AFREQ% -b 8 -e signed-integer -t s8 "%fname%.s8" "%fname%.wav"

@rem sox --multi-threaded --buffer 524288 --input-buffer 524288 -S -V6 -D -c 1 -r %AFREQ% -e float -b 32 -t f32 "%fname%.f32" -n spectrogram -z 96 -q 13 -w Kaiser -y 513 -x 496 -o "%fname%_32.png"
@rem sox --multi-threaded --buffer 524288 --input-buffer 524288 -S -V6 -D -c 1 -r %AFREQ% -e signed-integer -b 8 -t s8 "%fname%.s8" -n spectrogram -z 96 -q 13 -w Kaiser -y 513 -x 496 -o "%fname%_8.png"


@rem stereo
@sox --multi-threaded --buffer 524288 --input-buffer 524288 -S -V4 -D -G %file% -e float -b 32 -t f32 "%fname%.f32" rate -v -s -I %AFREQ% %aproc% stats -b 8 stat
@sox --multi-threaded --buffer 524288 --input-buffer 524288 -S -V6 -D -c 2 -r 48299 -e float -b 32 -t f32 "%fname%.f32" -c 2 -b 8 -e signed-integer -t s8 "%fname%.s8" %clip% dither -f %noiseshp% -p 8 stats -b 8 stat 
@sox --multi-threaded --buffer 524288 --input-buffer 524288 -S -V6 -D -c 2 -r %AFREQ% -b 8 -e signed-integer -t s8 "%fname%.s8" "%fname%.8svx"
@sox --multi-threaded --buffer 524288 --input-buffer 524288 -S -V6 -D -c 2 -r %AFREQ% -b 8 -e signed-integer -t s8 "%fname%.s8" "%fname%.wav"

@sox --multi-threaded --buffer 524288 --input-buffer 524288 -S -V6 -D -c 2 -r %AFREQ% -e float -b 32 -t f32 "%fname%.f32" -n remix - spectrogram -z 96 -q 13 -w Kaiser -y 513 -x 496 -o "%fname%_32.png"
@sox --multi-threaded --buffer 524288 --input-buffer 524288 -S -V6 -D -c 2 -r %AFREQ% -e signed-integer -b 8 -t s8 "%fname%.s8" -n remix - spectrogram -z 96 -q 13 -w Kaiser -y 513 -x 496 -o "%fname%_8.png"

@del -q "%fname%.f32"
@del -q "%fname%.s8"

@pause
@endlocal

Btw - i have no trusted normal audio files with useful audio content over 22..24kHz - some files i was able to find (Johann Sebastian Bach - Harpsichord Concertos BWV 1052, 1055, 1057 & 1058 Retrospect Ensemble / Matthew Halls, harpsichord & direction 24B/192kHz Stereo) seem to have high frequency content but for me it looks more like aliasing than real audio - anyway to be audible such signal (24kHz++) must have probably very high level and it will compete with noise shaping (as such signal level should be higher than -30 dBFS) - doubt if any natural instrument can produce such high level content but...

[pandy71]

Originally Posted by Gorf

There is still "dithering": add some noise at high resolution (16bit, 24bit) to reduce quantization noise at low resolution (8bit).It's a very old concept - Paper from 1964:https://ieeexplore.ieee.org/document/1088973More practical explanation:https://www.allaboutcircuits.com/tec...on-distortion/(What is Dithering? Using Dithering to Eliminate Quantization Distortion)

Concept is old but it was not well understand till 1990 (1990 is just convenient way to describe change in approach).But dither is not noise shaping - you can have dither without noise shaping, noise shaping without dither. primary goal for dither is quantizer linearization so quantization noise is no longer correlated with your signal.Noise shaping is way to push quantization noise out of interesting area - in case of human auditory system usually noise shaping address psychoacoustic characteristic - this is very interesting area.

[Gorf]

Originally Posted by pandy71

Concept is old but it was not well understand till 1990 (1990 is just convenient way to describe change in approach).But dither is not noise shaping - you can have dither without noise shaping, noise shaping without dither.

exactly
I did not say otherwise - in fact I brought it up as a method to reduce noise for a potential separate bass channel that is replayed via DMA at 28kHz, where noise shaping is no option.

[pandy71]

Originally Posted by Gorf

exactly I did not say otherwise - in fact I brought it up as a method to reduce noise for a potential separate bass channel that is replayed via DMA at 28kHz, where noise shaping is no option.

I just made clear separation as dither gain from oversampling is not so useful as gain from noise shaping - ideally dedicated application (sample player) for Amiga should optimally use Paula - as pointed by Vanderkooy and Lipschitz 4 times oversampled 8 bit audio with optimal noises shaping can deliver better results than audiophile oriented SACD format (technically this 64 times oversampled 1 bit bitstream so literally 1 bit DAC Sigma Delta output recorded on disc.

[Gorf]

Originally Posted by pandy71

I just made clear separation as dither gain from oversampling is not so useful as gain from noise shaping - ideally dedicated application (sample player) for Amiga should optimally use Paula - as pointed by Vanderkooy and Lipschitz 4 times oversampled 8 bit audio with optimal noises shaping can deliver better results than audiophile oriented SACD format (technically this 64 times oversampled 1 bit bitstream so literally 1 bit DAC Sigma Delta output recorded on disc.

That is true and 8bit with oversampling and noise shaping is probably the way to go, but I still think using the second channel per side (which would be otherwise unused) for the bass part of the source material might be a way to increase fidelity even further.

Even the not 100% temporal alignment of these channels are probably irrelevant if one of them played only the lower frequency part of the source material:



https://audioxpress.com/article/audi...ally-a-problem

While this site talks about jitter, the same would be true for audio-cannel misalignment

[saimo]

@pandy

Many thanks for the hard work!

One thing, though:

Originally Posted by pandy71

Due high level of ultrasonic signal (noise) i recommend to apply 20..22kHz low pass filter at the amplifier input or use moderate power level to not burn your tweeters (or use speakers insensitive to excessive high levels of ultrasonic signals)

Exposing the users' equiment to such risks and asking them to take this kind of precautions is a no go: I do understand the theoretical research objectives, but personally I'm after a practical solution to use in contexts like demos and games.
By the way, it's years that I no longer have speakers: I moved from place to place several times, and every time I had to get rid of everything, until I just grew tired of that. So now my A1200 is connected to a monitor with unsignificant speakers and for these tests I'm simply using commercial-quality headphones. Also, I tried to sample the A1200 output connecting the machine to the PC, but there's a consistent noise floor that would hamper any kind of software-based measurements.
In short: I'm not really the right person to make appropriate tests I hope my tools (which I have just updated and I'll upload after testing them on the A1200) will be of help to others, though.

[saimo]

Uploaded updated tools.Download & informationChanges: * removed padding bytes from the signal data in the AQA2 files (so, sAQA2 no longer supports the old AQA2 files); * added rounding to the 16->14-bit conversion code in cWAV16toAQA1; * made cWAV8toAQA2 accept signal files which contain enough data with respect to the carrier files (before, it required the data ratio to be exactly 2:1).The audio files were also updated, as I have re-rendered the WAVs from the original 24-bit FLAC (instead of the intermediate 16-bit WAV) and then converted them to AQAx with the updated tools.Now I'm going to make some tests by massively reducing the amplification of the test song and then by hearing the results on the Amiga.