The thing I loved about the Amiga sound was the creativity and technical abilities of the people who got around it's limitations. For instance, writing music around chord samples and careful re-use of voice channels between note & percussive sounds. Few musicians did this better than a Finnish composer who went by Di33y (check out Towards & Back and Banana Split, and all of their tracks are brilliant). I would've loved if a 'next-gen' Amiga sound were to just add a Motorola DSP like the Atari Falcon, NeXT workstations, some 90's Apples, and just let devs invent the soundchip they want per se. Was a huge fan of Roland and Yamaha XG MIDI artists who pushed the technical limitations of GS/XG in the day. One of my fave fromthe GS world was Anomaria and the in-house demo tune writers at Yamaha really pushed what XG could do in terms of effects buses and modulations. Recently bought an MU128 and have been getting back into playing with MIDI.
The Apollo FPGA core used in their V4SA and V4 Series Accelerator cards has a 16 bit audio version of Paula that is fully backwards compatible and has a bit more channels to play with.
right, just pick worst of the worst ever , if you can even call em compositions... and turn em into midi. like turning shit into diarrhea. just priceless
While the MT-32 sounds cleaner and brighter, Paula was more flexible. An MT-32 will always sound like an MT-32 since the sounds are fixed. Many artists (particularly in the demo and mod scene) were able to create amazing music using just Paula. Also, many early PC games used general MIDI or MT-32 even for sound effects, which could lead to some bizarre situations such as hearing what was obviously a snare drum play when you crashed your car in a racing game.
I agree with you, but wanted to note that the MT-32 could have new patches downloaded via MIDI SysEx messages. Most games that supported the MT-32 utilized this, with the exception of a few non-Sierra titles, like Dune 1 and Monkey Island 1.
@@RetroDawn True, although the fundamental architecture is still fairly limited (short attack samples and simplistic waveforms) where as Paula can play any length sample, stream new samples on the fly etc. and is only limited by the amount of RAM and storage space.
@@JimmiG84 100% agreed! I'll take Paula over MT-32 LA any day. More to the point, I'll take what the Europeans (including British & Irish ofc) did and do with the Paula over anything I've ever heard from the MT-32 (mostly USians and Canadians). And I say this as a USian, myself.
@@RetroDawnI‘m an Amiga kid, made music with the Amiga myself… and no, I don‘t take Paula over an MT32 for game music. There are very few games that sound better on Amiga than on MT32, especially with custom patches. The MT32 comes with higher bit depth, higher sampling rate, more voices AND integrated, controllable effects like reverb and delay. Of course there is no comparison to games mainly made for Amiga where the music on Paula was an art in itself. That‘s something completely different, thou. I would have loved to see far more Amiga games support MT32 or GM IN ADDITION to Paula sound. That would have been the king‘s solution.
I came here to make this exact statement. It is strange that NO improvements were ever made to the Amiga's sound capabilities. No additional voices, no higher fidelity (e.g. 12, 16 or 24 bit), no built-in special effects like echo/expand/contract/flange/etc. It was impressive when it was first released but the competition did not sit still. And with MIDI capability built-in to the Amiga's main competitor, the Atari ST, it's even more amazing that they didn't improve the sound for the next iteration. In an interview from several years ago, Glenn Keller, the designer of the Paula chip, mentioned that there was almost going to be 8 voices instead of 4, but it didn't materialize for some reason.
It's true the the Paula never got an upgrade and, yes it's true 4 channels suck, especially when you have to kill a main instrument to play sound effects (prime example for me was Pinball Dreams and Fantasies,) but some songs sound amazing with only 4 channels when they are programmed right. Chris huelsbeck, Alistair Brimble, Tim Follin all were amazing music creators.
The way the Amiga sound is generated in the Sierra games is ultimately a kind of MT32 emulator. If the individual tracks had been created individually as MODs, with the necessary samples, much more could have been achieved. But that would certainly not have been in Sierra's interest, because they distributed the MT32 back in the days. Nowadays you can get the MT32 sound quite cheaply at home with the PiMT32. Back then, it was simply unaffordable for the typical poor student.
Thank you for this video, really interesting. I am a big Amiga fan and I am a big Sierra fan as well, something that apparently is quite unique ;-) I played most of my Sierra games on my dad's PC and my dream was always to have a computer with a Soundblaster and MT-32, I think that was the ideal setup for Sierra games. To see Sierra games on an Amiga using the MT-32 is quite amazing. I often thought about this, but didn't know it was possible. I do think Sierra could have done a lot more with the Amiga sound. Rise of the Dragon is one of my favourite (Sierra) Amiga games and I think that sounds amazing and the MT-32 does not sound a whole lot better.
Paula was awesome, incredibly flexible and easy to use. Early Amigas were able to access 512K Chipmem that also was shared with graphics. The main program may also reside their (this was dependent on if you had extra memory above 512K and also the program size) needless to say memory space was limited and the original 512K set the standard (even when updates occurred expanding the Chipmem 1MB then 2MB) This limited how many and at what quality (cleaner samples require more memory and the Amiga was capable of rather clean samples) now its an easy task to swap samples out, but between laziness and not wanting to add disks (they cost money) it limited the quality. The fact is when the game loads it will contain music, video and program and there maybe some very limited memory. Some games did detect extra memory and utilize that by swapping both graphics and sound in and out of the Chipmem which allowed for better of both (as I said flexible). When the AGA chipset came out everyone was disappointed that Paula hadn't been upgraded (myself included) however with the Chipmem increase to 2MB Paula could now have much larger samples and sound much cleaner. The MT 32 was a instrument player and not for sound effects and while it was possible to change the instruments, doing so would require it to do so over a slow as hell interface (MIDI was not designed for mass data transfer) so (as far as I know) games did not update the instrument set while the game was running. While few native Amiga games supported MIDI, with an accelerated Amiga you can run ScummVM and have access to a whole bunch (whole bunch being the scientific term)
*The problem with the MT-32 was that the music* sounded more synthy and very similar to each other versus MODs. I'd rather have 4 (or later 8) voice MOD songs.
Fair enough! I also think that today’s demoscene productions really show how flexible the Paula is. However back in the day it seems there wasn’t the talent or time to push the Paula to its limits. Even the SID is pushed way past its limits nowadays. And between Paula and SID… I think I would pick the SID! :)
@@root42 This same problem with all PC games, they had to cut back to save disk space, and memory space, these games are on far too many floppy’s already, and the 512kb of CHIP ram was not a lot to work with, modern demos are made for 4 x chipram (2mb), this means you can have better quality sound and graphics.
@@kjetilhvalstrand1009 Not entirely true. Shadow of the Beast Games all ran on 512 OCS, Leander, Agony, etc all had excellent sound and music. It depends on the coder(s) and musicians.
you could not have much of 8 voices in games, also Amiga has terrible range. Another problem is that the ChipRAM is limited to 512KB and samples are hungry, so majority of games used 12KHz or even 6KHz samples to fit in. Also another factor is that if you play music with 4 voices, you dont have in game sounds. Paula also lacks any mixing control. It was a terrible sound chip made with cartridge based gaming console in mind, it become very obsolete fast and actually the Achilles heel for Amiga architecture.
@@root42 Yeah, I noticed this when I first tried the DOS ports of Monkey Island 2 and Indiana Jones & The Fate of Atlantis after enjoying those titles on my Amiga 500 back in the day. I can live with the reduced color palette in the Amiga versions but the missing animations, sound effects and music can be quite painful IMHO. You got me exited to hook up my Roland CM32L or MT32 to my Amiga computers and try out some of the tiltles you show but sadly I don't have the boxed versions, only the whdload versions 😞
@@Jivemaster2005 Yes, exactly. If you search around the web you will find disk images for those versions I showed. However WHDLoad might also have the install program included and/or the driver, which you can enable by editing the resource.cfg
Paula, in 1985 to 1990, was astounding compared to any competitor at the time. Also, I am amazed that Jack Tramiel allowed a midi port to be standard on the Atari ST, for such a port was for specific uses and Jack did not shed a dime for extras.
Interestingly, on the PC side, I had a SoundBlaster 16 which I paired with a Roland SCB-7 (I think. There was also the SCB-55 but I don't think I had that one). The music in my games with the Roland sounded pretty much how it sounds here on the MT-32.
I got an MT-32 for myself off EBay back in 2007 when the emulation wasn't good yet. It was well worth the price(about $30 plus shipping if I recall) since I got to hear all these old games how they were composed. I haven't taken it out in years, though...I should probably let go of it. The key advantages of it over Paula sound really came down to three things: reverb, polyphony, and envelopes. In theory all three could be achieved with a much faster CPU doing software synthesis and mixing, but most Amiga games in its commercial lifecycle barely scratched the surface of what Paula could do already - lots of "music or sfx, not both", reused Soundtracker samples, ported Atari ST assets, and generally not budgeting anything towards sound and focusing all the energy on screenshot material. That's the kind of thing that I think tended to define the Amiga most - it could always do more than the software aimed to accomplish. It didn't need more hardware power, it needed more software support and documentation to accommodate deadline-driven development.
The amiga when it came out had the best sound of any machine on the market, the problem was as well as with its graphics that they did not develop the machine fast enough further and by the time the PC had caught up and surpassed it in the graphics department, the sound department had as well. Funny thing is that the ST had a longer lifespan for many users despite having worse sound because the integrated midi port which was a low hanging fruit on top of the serial port pushed it straight into the musician scene which needed it to control their synthesizers. Btw you dont need a real MT32 nowadays to get hardware MT32 sound, the MT32Pi Project does the same job with some adapter heads!
Games, which pushed MT-32 on edge, were games from D.I.D, Inferno and TFX. They focused more on custom patches, which used virtual analogue part of its architecture. Biggest problem of MT-32 is, that it didn't inherit samples from D-50, but ROM was more focused on "realistic" instruments, which sounds pasticky today. :-)
The only thing that came close was the QS300 mode hidden in the Yamaha DB50XG and that was an oversight by Yamaha. It uses the same sound chip as the QS300 workstation and they forgot to disable the edit mode when they designed the DB50XG. Subsequent sound cards and modules using that chip kept only the TG300 and XG modes and disabled the QS300 mode.
@@root42 Well it did not age well, but it was the best affordable synth of its time for a short period of time, and if you want a perfect 80s synthi sound the mt32 is the one to use!
You could also connect most PC-Soundboards to the Amiga - with Soundboard I mean those MIDI-based-boards you could attach to the pinheaders ontop a Soundblaster and compatible cards. There were many, one of them was the Yamaha DB50XG, a board so famous you can still find people using it on TH-cam. The trick, all those boards just used TTL-based RS232 over the pinout. A simple Adaptorboard with some MAX232 was enough to connect them to the Amiga (and the ST and the C64 and whatever RS232 system you used). To behonest, the idea behind Soundboards was already dead in 1994 when the first cards with DSP arrived. At first those cards did a hell of a workaround to be MIDI-compatible, hiding the RAM and the DSP behind a MIDI-Interface. But the writing was on the wall, in 1996 the first cards with DSP but without MIDI-hardware-layer showed up and with AC97 sound MIDI became totally irrelevant for PC-based audio. Also with the arrival of MMX the DSP was pretty fast dying off too because a 200Mhz MMX CPU could easily mix 100 audio channels at 100khz and 24 Bits.
AC97 is a jittery, clippy, low resolution nightmare though, sound awful to this day, as does using an older CPU as a DSP, noisy, high-latency, poor timings, a sound card and hardware DSP + DAC has always been vital for HQ audio. Unless you are just using low quality speakers and virtually no AMP power.
@@Wobble2007 You are treating in a field where you are guessing and not knowing. While AC97 is limited to 16Bit, Stereo and 48kHz this is still even slightly above CD-Quality. As long as a good DAC is build into the system it is pretty much CD quality. A DSP has ABSOLUTELY NOTHING to do with audio quality. It is just a linear processor for digital data. And MMX is doing EXACTLY the same and was basically developed to replace the expensive and driver-problematic DSPs back then. Funnily HDA nowadays has an minimalistic Mixer-DSP included again which e.g. Linux and ASIO drivers for real time critical audio simply doesn't use because it is slower than they CPU. The Audio-Output even for pre-AC97 was already DMA-based for over ten years and therefore CYCLE-precise down to the Nano-Second. Even my Amiga 1000 from 1985 used DMA with a precision of around 150ns. That is 10000 times lower than anything a human could make out. And latency only depends on the buffer size. About the quality of AC97: It depends on the DAC. A cheap SiS-board usually had a crappy DAC, but at least the AC97 audio on my AMD Irongate board had studio quality (at least by 1999 standards) and a noise level lower than most CD players. I remember that Tomshardware or Anandtech tested some AC97 solutions back in 2000 and they even beat some dedicated expensive sound cards.
@@CrassSpektakel I have yet to hear acceptable AC97, you will get missing data in CD audio playback on a AC97 CODEC IC, something as simple as guitar scratch will be auditorily missing, as will breath and other subtle tones, due to low resolution. A DSP can mean everything as far as audio quality, it allows for EQ, reverb, flanging, attack, chorus, and so on, it can really help lower resolution audio, take the crystallizer on the X-Fi series for instance, it can work miracles. Latency depends on a whole lot more than buffer size, drivers make a big difference for one, ASIO or WASAPI for one thing will always trump DirectSound. DAC quality varies wildly, you will never find an onboard DAC that sounds good outside of Apple hardware, who at least use a reasonable quality default DAC, PC hardware always uses awful CODEC based IC's.
@@Wobble2007Everything you claim a DSP can do a CPU can do too. With MMX even faster. Even a PentiumMMX@166Mhz outperforms everything the back then Motorola 56k DSPs could do. Well, the DSPs needed less power for the same results but a modern Ryzen CPU can do complex mixing and converting of thousands (!!!) of channels in real time with less than 1% of its CPU power. In fact in the 1990ths I wrote a software-audio-mixer for the Amiga which did all of this in real time on an 68000@7Mhz - 12 stereo voices in 14Bit audio at 56kHz. DSPs are just niche co-processors running DSP software. You can just write the same software for another plattform. It is not magic. Not even smart hardware by todays standards. Back then I used the AC97 sound of my AMD Irongate as a software defined modem for amateur radio (AC97 was also known as MC97 - Modem Codec 97). Why? Because my oscilloscope showed a timing precision of
In your video at position 8:05 you can see that the lower left pin of the serial interface is pushed inside. If I were you, I would repair that before it causes a short inside. With a bit of luck, you can carefully pull it out with a thin needle-nose plier.
An external MIDI device would have been really nice for other sorts of games. How many games had to choose between music or sound effects? MIDI for music and Paula for sound effects would have been the best of both.
MT-32 also was flexible, you could upload new patches and create new sounds. But yeah, the Paula was basically a simpler version of for example the GF1.
You also only had 4 channels. And a low rate. And it's sample-based synth equiv which means it has that weird sound to it when you try to stretch a sample over too much frequency range (octaves). I'd take the MT-32 over Paula every time.
@@104d_3rr0r_vince Ohh I know. I'm from that era. I made plenty of mods and cut my teeth on trackers as well. I also know all the limitations as well (and the tricks that try to hide those limitations).
@@root42 you need to take into account when the Amiga was released and when sound blasters became the norm, there is a big gap. The MT32 came out quite a bit later, and you could equally add an AKI sampler and blow it out the water, what made the Roland MT32 good was game compatibility. Monkey island is a good example of how the Amiga was better than your average PC at the time.
It sounds like clipping, the sound cuts of at high amplitude, it also sound like its playing, low quality samples, without mixing. This games where made for 7Mhz 68000 nothing more powerful.
Yeah - those pops at the beginning of every note (in most examples)? That's about what happens if you don't start your samples at a zero-crossing - it's almost like they didn't really have a clue about how to use something sample-based like the Amiga, AT ALL (which doesn't make ONE bit of sense to me, as that should be one of the most basic things anybody worth their salt working with audio in any way learns pretty fast... I'm pretty much an absolute dilettante in that regard, but even I know that much, so that kind of error on absolute pros doesn't make one bit of sense 🤷♂...)
Very cool idea for a video. It always puzzles me, why you only have 7.000 subs. I also wonder why they did not sample the instruments from some professional synthesizer for the amigasnd.drv. Also interesting would be to see a side by side comparison on the graphics amiga vs. atari st vs. vga....
Its certainly an interesting comparison - but i cant help but feel they might have been a bit lazy with the conversion. Some of the short throwaway jingles could have been a straight up sample of the equivalent MT32 ones. Another piece which has a single pipe playing has a nice reverb on the MT32 which could have been achieved easily on the amiga by playing the sample notes across multiple channels so the reverb doesnt get cut off from the previous note when the next one is played.
Yeah none of these examples showcase the amiga's full audio potential! You only have to listen to an MT-32 playing the GODS or Speedball 2 music to hear how great samples can be. Also people loved using the inbuilt MT-32 reverb a lot and it hasn't aged well.
yea this is apples and oranges.. not better. different.. personally as a musician i'd take a sampler over an fm synth anytime.. th-cam.com/video/Pu0W2FDdMsM/w-d-xo.html image FM synth doing this song?
@@8bitwidgets Not FM, its linear arithmetic synthesis, which consists of a short ROM type sample attack and then a subtractive (think VA type in nature) phase after the initial attack portion. FM was actually on patent by Yamaha
@@jasonblazgk9973 ah LA synthesis.. i had a GR-50 that had that.. still very different tech and capability. paula lacks the built in synth features.. but i enjoy the 4 dedicated voices of it.. but in terms of music in games the richer synth capabilities of LA has it's place, but as a musician i still favor paula. i didn't play many games on the Amiga.
considering how old it is it sounds amazing i didn't think a computer from that year could sound so good seems like it was made in 1990 .seems like back then support for hardware addons was rare a lot of people had a dos pc but the sound on it is the most basic sound possible so this was a huge upgrade for pc owners. but on the amiga this would only be a minor upgrade but even other systems with more limited sound like the st and zx spectrum and so on didn't have any games that supported this or similar upgrades even on pc it took many years for it to get support from most games . but maybe it was just because it took up more memory to have different versions of each song
As you say, audio on the Amiga was affordable and good but not perfect. Like so many things the tight integration with the video timing made an hardware upgrade to Paula worryingly complicated. But then you could bypass a lot of limitations by using the CPU for mixing and clever combining of channels so all in all even an Amiga 500 was able to output 14Bit stereo at 56khz by pulling every trick available and it didn't even cost any additional CPU power. But that was just "one" channel. Sure, you could mix channels in software but that was where a stock 68000 was just out of its league, barely getting along with eight voices. With my 68030@25Mhz though I was able to get ~20 voices, although at high CPU load.
Do you are anyone know about the studio mod for the MT-32? People would pay for a sound mod , I think a firmware mod/changing output ports, but am not sure. I did see old listings for it and people asking about it ,but it was unclear what the mod was , only that it had poor sound when it came to high-end hifi setups/studios, and you could fix this with the mod.
I haven’t watched yet, but as a fairly respected Amiga musician and also the composer behind a popular MT32 video, my hot take is that this is a flawed premise 😁
@@root42turns out it *was* a flawed premise I’m afraid. It lacked in context and scope, with some inaccuracies and missed nuance. I’m not trying to be mean btw.
Großer Fan von MT-32 hier! :D Wenn du überlegst, daß der Roland D-50, also der LA-Synthese-Synthesizer, der quasi hinter dem MT32 steckt (dann natürlich als ausgewachsener Synth, MT32 ist halt abgespecktes Soundset) Ende der 80er 4000 (!) Mark gekostet hat wirds ein Schnäppchen ;). Die LA-Synthese ist eh clever gewesen, Mini-Sample meets substraktive Synthese, zwar digital aber abgeleitet von den analogen Synthies. Im Vergleich zu FM wie bei Yamahas OPLs (die im Yamaha DX7 als revolutionären Mega-Seller-Synthie, der den Pop-Sound der 80er definierte) um einiges einfacher zu verstehen. Wenn nicht Roland in Sachen Menü-Führung bis heute grausam wäre ;). Spätere Midi-Soundboxen von Yahama (XG Standard) find ich persönlich noch nicer (die Filter sind teilweise episch), aber da war der Amiga ja schon tot.
On a module like that it's more likely that you connect both midi in and outs to be able to dump sysex data to it. Seems like it does that when the game starts up. I can't really see any other use for the midi out otherwise :). If you want to connect devices in a chain you'd use the midi thru port :)
The MT-32 is a lot nicer on the ear here as the audio has a greater dynamic range/bit depth, the MT-32's DAC is likely a bit nicer than the amiga and the channel multiplexing is more sophisticated. Which I guess we should kind of expect from a box that was more expensive than the amiga500 and sold as a (not quite) pro synth. Though I understand there are bugs in the implementation if the MT-32's DAC. I guess the main drawback for the MT-32 is the fixed palette of sounds which the amiga doesn't have The amiga is certainly much more versatile as it can be programmed to do all sorts; PCM playback, FM synthesis, even rudimentary wavetable synthesis (though I doubt the latter two were ever used in any games). The only thing that has always bugged me is the terrible channel multiplexing, having two channels that are panned hard left and two that are panned hard right isn't great and means the mono playback is mostly the only sensible option. You do see committed mod writers emulating stereo effects by using the same sample in the right and left channels but again not common in game music. I was always surprised that this, of all things, didn't get addressed in the shift from OCS to AGA. I wonder how the amiga music examples in the vid was created? If they essentially just re-sampled sounds from other better synths/soundcards, even with the 8-bit limitation if they were somewhat lazy about it that could contribute to the lack of dynamics both in the instruments and the richness of the sound
Nice video! You should keep in mind though that you are comparing the bare Amiga to the MT-32 with Reverb enabled. This is not really a fair comparison imo.
I know -- this is a bit of an unfair comparison. The Sierra ports are also notoriously bad, compared to what was actually possible on the Amiga. However it shows what we often got, versus what money COULD buy at the time.
It has nothing to do with the Engine. Hero's Quest (Quest for Glory EGA version) with the SCI0 engine perfectly supports the MT32 on the PC because the driver is included. I'm not sure how the Amiga "detects" that you want to output the audio using the MT-32 instead of "FM" or whatever the option would be for the Amiga. Now I'm curious how it would work with the AtariST ...
On the ST those games had MT-32 drivers included, IIRC. But I think the early SCI0 games didn't have the full MIDI data on the Amiga. But I could be wrong. I didn't get any of the MT32 drivers from other games to work with earlier ones.
Ah, here is the SCI Wiki's take on the sound resources: sciwiki.sierrahelp.com/index.php/SCI0_Sound_Resource_Format At the very least the Amiga games have the bank.001 resource, which contains the samples. One would need to check if the resource files also contain the patches for MT-32, or not. If they are missing this would not work, no matter if you had a working MT-32 driver or not. I need to check that with some SCI utilities. It would be nice to activate more SCI0 games to play MT-32 music, though!
Sounds like they recorded samples off the mt32 and stored them as PCM samples for the Amiga to play. The mt32 is just an FM synth, which the Amiga doesn't have. It only does 4 channel of PCM audio, which I would argue is better, aside from this particular case where Sierra just did a rather nasty port.
"Better" is in the eye (and ears) of the beholder. The amount of equalizers and DSP's all around should be proof of that. If i push Turrican II through WinUAE to my headphones, the output is crisp and... nerve-wrackingly so. But if i push it though my Yamaha YST-M15 plus YST-MSW10, ehhhh, almost sounds the same as the Amiga did, which is how i liked it. Similar thoughts would be me missing WinAMP's Enchancer DSP because it was the only thing that made it "sound right", and cheap Sony headphones because they "sounded right" out of the box. Different yes, better depends...
I'm a USian, but I've always *far* preferred the game and demo music composed and produced by the Europeans on the non-PC computer platforms compared to that of the USians for any platform, including PC and MT-32. Edit: As a USian, I include British & Irish as Europeans, ofc. I say this b/c I know that at least some Brits (and perhaps a small number of Irish) don't identify as European.
Little better midi player than MT32 was ProjectXG, it was based on Yamaha DB50XG daughterboard (SW60XG was standalone card, CS1x standalone synth). XG standard is far more powerful than General MIDI or Roland GS, good example is XG midi music from Final Fantasy VII. But i love Amiga's Paula as is, just 4 channel pcm sample player without real synthesis, trackers which can do synthsouds are probably precalculated one cycle waveforms, varied with some realtime commands.
After all these years I must say, the MIDI-sound system does not sound as good as I remember, back then it was spendid, when I listen to it now it is a bit shallow, plastic and cheap. When MP3 came to computing music the MIDI system lost its spender, vividness and excitement. The Amiga sound chip could play back samples and back then it was quite an achievement, now only real sounds sound real.
I wonder why the samples are clicking so much. Is it because samples do not start from zero crossing or something else? I've never owned Amiga so I don't know how common this is.
@dr.benway1892 Yes, that's roughly how most of the examples in this video sound like - just about the worst I've ever heard from an Amiga... But as I learned from some comments around here, Sierra was _distributing_ the MT-32 at the time, so they may have had kind of a vested interest in not making the 'competition' sound 'good' in any way, shape or form, so 🤷♂️😁...
ok to be honest, if you want to hear the real capabilities of the Paula chipset you should play amiga games and not dos games created with adventure engine that are interpreted by the engine itself and not optimized for the host machine.. do not misunderstand me, i love dos game and also lucas and sierra games but their games were create mainly for pc dos computer that has no sound capabilites at all (until sound blaster arrive) and so when they were ported to the amiga they were ported "as is" without any kind of optimisation... it is nice to have midi on the amiga (bars and pipes and many others) but i do prefer the amiga sound...
Of course. Or just taking any of the great demo productions shows the power of the Paula. This is more of an example how little love some of the Amiga ports got.
WHAT IF SIERRA PROGRAMMED MUSIC & GRAPHX BETTER ? To be honest, MT-32 was very expensive and it sounded somewhat flat anyway. Way better music got made on native Amiga sound hardware by musicians like Chris Huelsbeck, Jeroen Tel or the Psygnosis people using samples, or on the Megadrive's Yamaha chip by musicians like Yuzo Koshiro. Sierra's artists were just lazy asuming that everybody had expensive hardware, then grossly underutilizing "standard" capabilities such as EGA and Soundblaster, just because they didn't match their development set-ups. As a counter-example look at what Rare did for Nintendo on the SNES with Donkey Kong Country, which looks and sounds awesome on such underpowered hardware (which did sound mainly by sampling, just like Paula).
I've heard much better sound from the Amiga in other games. LucasArts games especially sounded great. Sierra games on Amiga were known for sound that was lacking.
Paula chip frequencies are a bit tricky to measure. Isn't it so that if you want a simple square wave you already need 2 samples and if these are in 28Khz the actual frequency would be closer to 14Khz. But yeah, don't get me wrong, despite a bit weak bass tones and saturated mid-tones & limited hi frequencies, Paula gives the "Amiga" sound. And I hear people complaining about "having only 4 channels"... Get on with it! If you can't put up a good song with three sound channels, study first with only two!
There are awesome tunes using the TED (two channels), TIA (two channels, weird notes) and even the original Speccy and PET (one channel square wave). Indeed much of the creativity stems from those limitations!
@@root42 Just recently attended a seminaire about Speccy's sound chip and the evolution of the model (at Skrolli-party). Amazing improvements from fairly horrible one-channel beeper to CPU-hog 2-channel mode and towards newer 3 channel chip and further into filter-like "synth-trickery". BTW, Cool video and sweet shirt :D
The only really big flaw with Amiga's audio was the hard panning. 3 channels for music and 1 for sound effects would have been sufficient but sound effects only coming out of one speaker sounded awful. And balancing three channel music with two on one channel and one on the other was painful. You were best off just externally mixing the two channels, but most games didn't design for that and instead you got this norm where you got music or sound effects, often you could choose which but rarely did you get them at the same time.
I listened to music of unreal games and games like crusader lately or even the music of the unreal demo from 1992. I also like the music of pinball games on the amiga and pc. I prefer listening to music of impossible mission or shadow of the beast instead of midi music of pc games. It all would't exist if an amiga 'pushed' developers to create midi files without sample content. I think the history is just fine 😇 not to be negative in any way. I love what you did in this video 👍 the latest music I made is actually a general midi file for a doom mod 😂
Well, even when used with the most expensive 14" CRT TV 2 inch mono speakers of Sony/Panasonic etc you won't be able to tell the difference between 28khz 8bit Amiga vs 44.1khz 16bit CD quality audio of a console so there's that......
Amiga did not play synthetic sound, it played prerecorded sound, it technically can have been generated but 68000 7mhz was not powerful, it takes lot time to do so, it be lot larger technical challenge to convert soft sounds into wave sounds. Then having the sound prerecorded, my guess is that sampled sound low frequency, to save space.
@@DS-pk4eh Sound is rage of 20 hertz to around 20 000 khz, cd quality is 44.1 Khz, now that’s far away from 7mhz, so should have problem doing it, but it will take a substantial, amount of CPU resources. This resources you want to use on the game.
@@kjetilhvalstrand1009 That is not how that works. The 68000Mhz has nothing to do if it is capable to produce the sounds human can hear. Anyway, I was thinking how big chip like that from Roland would be to have it next to others chip in Amiga. So, unless you know the size of the chip in Roland and how it worked, it will not answer my question.
I never understood the MT32; it made everything sound the same. only later, with wavetable and XG this became mature. I never really liked the sound of the Roland stuff... except for the TRx0x and TB303 of course ;-)
GUS series cards don't have reverb, they can sound pretty flat and boring, so you have to add some reverb during sample edit. Advantage is, you can use any sample.
If the Amiga had come out with better sound, it would still have sounded better than its price competitors assuming same price. But really this would have cost far more and so people wouldn't have bought it. They'd have bought a sampler and Atari ST instead for the same money. If they had wanted to make CD quality audio. And this is why the Amiga didn't come with better sound than the Amiga. Ever. You can do a lot with software for free, hardware costs $$$.
Boar, die Amiga-Portierungen spielen gesampelte midi Samples... Und das ist mal schlecht konvertiert! Ich finde, das kann man nicht vergleichen! Ein mies digitalisierter midi-Sample klingt neben einer General midi Box natuerlich schlecht! Hier sollten Scene Musikstuecke verglichen werden, nicht schlechte Audio portierungen.
This sounds very bad, like a cheap Casio, and the music (as in the notes) is also very poor and stale. I'm trying to bend my mind around the concept that someone could actually think that simple-minded barren songs using terrible instruments sounds better than all the bangers in hundreds of Amiga games, and thousands of demos. The Amiga sound, and not the cheap sound canvas sound, is largely what sold millions of Amigas.
@@madigorfkgoogle9349 The first Amiga tracker was designed by Karsten Obarski for making in-game music. It was cloned to make other trackers like NoiseTracker and Protracker. Whether writing tracker music for demos or games, it's the same thing. Note: The Amiga can sum channels to play back 14-bit stereo samples. Not bad for a 1985 machine and puts it ahead of the Ensoniq Mirage and almost to the level of the original EPS (minus the polyphony).
The game conversion team at at Sierra simply did not in anyway have the skill set to leverage the strengths and unique ability's of the Amiga's hardware those that did knew exactly how capable the Amiga,s sound hardware was capable of. th-cam.com/video/BB3_33HFyus/w-d-xo.html
Używam amigi od lat 90 a nie wiedziałem że na miga pisano gry które potrafiły wykorzystać port Midi z obsługą MT 32 w Polsce się o tym nigdy nie mówiło Przynajmniej ja nigdy nie słyszałem a kupowałem co miesiąc miesięczniki Amiga i inne Commodore&Amiga
when you look at the poor quality of the games it doesnt need any better sound to be honest , its a shame it didnt get any 3d chips they could have put a better chip in like an s3 or something with more video ram that would have been great and pc games could have been ported on it i know people have done pc ports to amiga on aga and there pretty good but it would have been great for faster 3d then they could have had 16bit sound then it would have been awesome. thats why soon as the PlayStation 1 came out booom end of the amiga for me.
Agreed. The Paula can do better. But I think back then the know how wasn’t quite there yet. And yes, the MT32 is a consumer grade synth. It obviously sounds… synthetic. But back in 1989 Sierra games sounded better on the MT-32 than on the Paula, I think. Today demoscene prods show what the Paula is actually capable of.
@@root42 not really, Paula was a very outdated chip very fast. It was made with cartridge based game console in mind, so if you want to use in game music you have to resort to 12KHz or even 6KHz samples, also Paula has zero mixing control. The demoscene is usually using all the power of computer, using CPU for downmixing, you cant have this in games, especially not with higher sample rate. The ChipRAM is also a very limiting factor, since games had to run on lowest denominator ant that was 512KB ChipRAM Agnus. You cant fit much in that, on cartridge based console this would be no problem since you would fast switch those samples, but Amiga was different at the end... Also dont forget that these games were made in early years and many tricks didnt exist back then...
@@root42 I think it's also a matter of where you come from. In 1989 I was just getting into demos and was amazed by the music disks from Mahoney & Kaktus. The resolution-limited (grainy) sound actually worked great for me. When I first heard midi tunes they were very underwhelming as they didn't sound much better than those early Soundtracker modules and the demo musicians had vastly improved. Even some far more limited (technically) C64 SID tunes sound better to me. It's all in the mastery of the musician. If you listen to 'Guitar slinger' from 1993 you might mistake it for a 'real' tune on FM radio or an mp3.
even with Paula and 4 channels/music mods , most game-music really sounded great at that time, like risky woods, turrican2, lotus 3, pinball(s) ..etc
RIsky Woods? OMG. I know that game from DOS... Sadly, didn't have an Amiga, just a 286 at that time.
You probably already know about this, but the MT-32 Pi is actually a really decent (and a LOT more attainable) alternative. Sounds freaking great too!
The thing I loved about the Amiga sound was the creativity and technical abilities of the people who got around it's limitations. For instance, writing music around chord samples and careful re-use of voice channels between note & percussive sounds. Few musicians did this better than a Finnish composer who went by Di33y (check out Towards & Back and Banana Split, and all of their tracks are brilliant). I would've loved if a 'next-gen' Amiga sound were to just add a Motorola DSP like the Atari Falcon, NeXT workstations, some 90's Apples, and just let devs invent the soundchip they want per se.
Was a huge fan of Roland and Yamaha XG MIDI artists who pushed the technical limitations of GS/XG in the day. One of my fave fromthe GS world was Anomaria and the in-house demo tune writers at Yamaha really pushed what XG could do in terms of effects buses and modulations. Recently bought an MU128 and have been getting back into playing with MIDI.
The Apollo FPGA core used in their V4SA and V4 Series Accelerator cards has a 16 bit audio version of Paula that is fully backwards compatible and has a bit more channels to play with.
Wow, day and night! What a difference. More depth and volume.
This is my nerdiest thing of the quarter thingy. Thanks! 🎉
right, just pick worst of the worst ever , if you can even call em compositions... and turn em into midi. like turning shit into diarrhea. just priceless
Thank you very much for this video. This is the first time ever, that I see a video talking about the Amiga and MT32
While the MT-32 sounds cleaner and brighter, Paula was more flexible. An MT-32 will always sound like an MT-32 since the sounds are fixed. Many artists (particularly in the demo and mod scene) were able to create amazing music using just Paula. Also, many early PC games used general MIDI or MT-32 even for sound effects, which could lead to some bizarre situations such as hearing what was obviously a snare drum play when you crashed your car in a racing game.
I agree with you, but wanted to note that the MT-32 could have new patches downloaded via MIDI SysEx messages. Most games that supported the MT-32 utilized this, with the exception of a few non-Sierra titles, like Dune 1 and Monkey Island 1.
@@RetroDawn True, although the fundamental architecture is still fairly limited (short attack samples and simplistic waveforms) where as Paula can play any length sample, stream new samples on the fly etc. and is only limited by the amount of RAM and storage space.
@@JimmiG84 100% agreed! I'll take Paula over MT-32 LA any day. More to the point, I'll take what the Europeans (including British & Irish ofc) did and do with the Paula over anything I've ever heard from the MT-32 (mostly USians and Canadians). And I say this as a USian, myself.
@@RetroDawnI‘m an Amiga kid, made music with the Amiga myself… and no, I don‘t take Paula over an MT32 for game music. There are very few games that sound better on Amiga than on MT32, especially with custom patches. The MT32 comes with higher bit depth, higher sampling rate, more voices AND integrated, controllable effects like reverb and delay. Of course there is no comparison to games mainly made for Amiga where the music on Paula was an art in itself. That‘s something completely different, thou. I would have loved to see far more Amiga games support MT32 or GM IN ADDITION to Paula sound. That would have been the king‘s solution.
Paula's sound is so warm , nice to have midi too ✌️😎
The Paula was one of the few custom chips, that never got any improvements.
yes I hoped for 8 or even 16 channels on the A1200 and i was a bit disappointed
@@bernymeyer1578
And should have been upgraded to 16bit 44.1khz.
I came here to make this exact statement. It is strange that NO improvements were ever made to the Amiga's sound capabilities. No additional voices, no higher fidelity (e.g. 12, 16 or 24 bit), no built-in special effects like echo/expand/contract/flange/etc. It was impressive when it was first released but the competition did not sit still. And with MIDI capability built-in to the Amiga's main competitor, the Atari ST, it's even more amazing that they didn't improve the sound for the next iteration. In an interview from several years ago, Glenn Keller, the designer of the Paula chip, mentioned that there was almost going to be 8 voices instead of 4, but it didn't materialize for some reason.
Yes! exactly! Why it never received any useful upgrades is so beyond me! Even the GF1 was eventually upgraded by the Interwave.
It's true the the Paula never got an upgrade and, yes it's true 4 channels suck, especially when you have to kill a main instrument to play sound effects (prime example for me was Pinball Dreams and Fantasies,) but some songs sound amazing with only 4 channels when they are programmed right. Chris huelsbeck, Alistair Brimble, Tim Follin all were amazing music creators.
MIDI ports are basically serial ports with a different pinout all you need is a dongle.
A MIDI interface. I had one from Datel.
The way the Amiga sound is generated in the Sierra games is ultimately a kind of MT32 emulator. If the individual tracks had been created individually as MODs, with the necessary samples, much more could have been achieved. But that would certainly not have been in Sierra's interest, because they distributed the MT32 back in the days.
Nowadays you can get the MT32 sound quite cheaply at home with the PiMT32. Back then, it was simply unaffordable for the typical poor student.
Thank you for this video, really interesting. I am a big Amiga fan and I am a big Sierra fan as well, something that apparently is quite unique ;-) I played most of my Sierra games on my dad's PC and my dream was always to have a computer with a Soundblaster and MT-32, I think that was the ideal setup for Sierra games. To see Sierra games on an Amiga using the MT-32 is quite amazing. I often thought about this, but didn't know it was possible. I do think Sierra could have done a lot more with the Amiga sound. Rise of the Dragon is one of my favourite (Sierra) Amiga games and I think that sounds amazing and the MT-32 does not sound a whole lot better.
The most interesting fact is that Roland MT-32 price in 1987 was practically the same as Amiga 500. $695 vs. $699.
Paula was awesome, incredibly flexible and easy to use. Early Amigas were able to access 512K Chipmem that also was shared with graphics. The main program may also reside their (this was dependent on if you had extra memory above 512K and also the program size) needless to say memory space was limited and the original 512K set the standard (even when updates occurred expanding the Chipmem 1MB then 2MB) This limited how many and at what quality (cleaner samples require more memory and the Amiga was capable of rather clean samples) now its an easy task to swap samples out, but between laziness and not wanting to add disks (they cost money) it limited the quality. The fact is when the game loads it will contain music, video and program and there maybe some very limited memory. Some games did detect extra memory and utilize that by swapping both graphics and sound in and out of the Chipmem which allowed for better of both (as I said flexible). When the AGA chipset came out everyone was disappointed that Paula hadn't been upgraded (myself included) however with the Chipmem increase to 2MB Paula could now have much larger samples and sound much cleaner.
The MT 32 was a instrument player and not for sound effects and while it was possible to change the instruments, doing so would require it to do so over a slow as hell interface (MIDI was not designed for mass data transfer) so (as far as I know) games did not update the instrument set while the game was running.
While few native Amiga games supported MIDI, with an accelerated Amiga you can run ScummVM and have access to a whole bunch (whole bunch being the scientific term)
*The problem with the MT-32 was that the music*
sounded more synthy and very similar to each other
versus MODs. I'd rather have 4 (or later 8) voice MOD songs.
Fair enough! I also think that today’s demoscene productions really show how flexible the Paula is. However back in the day it seems there wasn’t the talent or time to push the Paula to its limits. Even the SID is pushed way past its limits nowadays. And between Paula and SID… I think I would pick the SID! :)
@@root42 This same problem with all PC games, they had to cut back to save disk space, and memory space, these games are on far too many floppy’s already, and the 512kb of CHIP ram was not a lot to work with, modern demos are made for 4 x chipram (2mb), this means you can have better quality sound and graphics.
@@kjetilhvalstrand1009 Not entirely true. Shadow of the Beast Games all ran on 512 OCS, Leander, Agony, etc all had excellent sound and music. It depends on the coder(s) and musicians.
@@root42 Both Paula and SID both have their own unique charm. I'll have both!!
you could not have much of 8 voices in games, also Amiga has terrible range. Another problem is that the ChipRAM is limited to 512KB and samples are hungry, so majority of games used 12KHz or even 6KHz samples to fit in. Also another factor is that if you play music with 4 voices, you dont have in game sounds. Paula also lacks any mixing control. It was a terrible sound chip made with cartridge based gaming console in mind, it become very obsolete fast and actually the Achilles heel for Amiga architecture.
This is awesome! Now we need this working in the Lucasarts games. Too bad the ones you show doesn't work as whdload versions
LucasArts would require a reimplementation. Would be nice as the original ports are severely lacking compared to the PC version.
@@root42 Yeah, I noticed this when I first tried the DOS ports of Monkey Island 2 and Indiana Jones & The Fate of Atlantis after enjoying those titles on my Amiga 500 back in the day. I can live with the reduced color palette in the Amiga versions but the missing animations, sound effects and music can be quite painful IMHO. You got me exited to hook up my Roland CM32L or MT32 to my Amiga computers and try out some of the tiltles you show but sadly I don't have the boxed versions, only the whdload versions 😞
@@Jivemaster2005 Yes, exactly. If you search around the web you will find disk images for those versions I showed. However WHDLoad might also have the install program included and/or the driver, which you can enable by editing the resource.cfg
The term that you are looking for is called the Nyquist theorem.
Paula, in 1985 to 1990, was astounding compared to any competitor at the time. Also, I am amazed that Jack Tramiel allowed a midi port to be standard on the Atari ST, for such a port was for specific uses and Jack did not shed a dime for extras.
Paula is the best girl. Pc music was horrible back then with very few exceptions.
This black look is crazy !!!
Interestingly, on the PC side, I had a SoundBlaster 16 which I paired with a Roland SCB-7 (I think. There was also the SCB-55 but I don't think I had that one). The music in my games with the Roland sounded pretty much how it sounds here on the MT-32.
I got an MT-32 for myself off EBay back in 2007 when the emulation wasn't good yet. It was well worth the price(about $30 plus shipping if I recall) since I got to hear all these old games how they were composed. I haven't taken it out in years, though...I should probably let go of it.
The key advantages of it over Paula sound really came down to three things: reverb, polyphony, and envelopes. In theory all three could be achieved with a much faster CPU doing software synthesis and mixing, but most Amiga games in its commercial lifecycle barely scratched the surface of what Paula could do already - lots of "music or sfx, not both", reused Soundtracker samples, ported Atari ST assets, and generally not budgeting anything towards sound and focusing all the energy on screenshot material. That's the kind of thing that I think tended to define the Amiga most - it could always do more than the software aimed to accomplish. It didn't need more hardware power, it needed more software support and documentation to accommodate deadline-driven development.
Definitely. The Paula‘s potential was only unlocked decades later, for the most part.
The amiga when it came out had the best sound of any machine on the market, the problem was as well as with its graphics that they did not develop the machine fast enough further and by the time the PC had caught up and surpassed it in the graphics department, the sound department had as well. Funny thing is that the ST had a longer lifespan for many users despite having worse sound because the integrated midi port which was a low hanging fruit on top of the serial port pushed it straight into the musician scene which needed it to control their synthesizers. Btw you dont need a real MT32 nowadays to get hardware MT32 sound, the MT32Pi Project does the same job with some adapter heads!
I had an MT-32 module that used to get swapped between my PC and my Amiga's, my Roland D50 set to General MIDI sounded the best though.
Games, which pushed MT-32 on edge, were games from D.I.D, Inferno and TFX. They focused more on custom patches, which used virtual analogue part of its architecture. Biggest problem of MT-32 is, that it didn't inherit samples from D-50, but ROM was more focused on "realistic" instruments, which sounds pasticky today. :-)
Inferno had one of THE best soundtracks ever. Amazing game in that regard. And yes, I agree that the MT-32 sounds a bit cheesy. But still I love it.
The only thing that came close was the QS300 mode hidden in the Yamaha DB50XG and that was an oversight by Yamaha. It uses the same sound chip as the QS300 workstation and they forgot to disable the edit mode when they designed the DB50XG. Subsequent sound cards and modules using that chip kept only the TG300 and XG modes and disabled the QS300 mode.
@@root42 Well it did not age well, but it was the best affordable synth of its time for a short period of time, and if you want a perfect 80s synthi sound the mt32 is the one to use!
@@root42 despite of everything I like more Mt-32 then SC-55, as SC-55 I consider as plain ROMpler.
@@atomicskull6405 MU100r could have such a secret mode. 😀
The soundchip was awesome as is - in chip music and sampled.
You could also connect most PC-Soundboards to the Amiga - with Soundboard I mean those MIDI-based-boards you could attach to the pinheaders ontop a Soundblaster and compatible cards. There were many, one of them was the Yamaha DB50XG, a board so famous you can still find people using it on TH-cam. The trick, all those boards just used TTL-based RS232 over the pinout. A simple Adaptorboard with some MAX232 was enough to connect them to the Amiga (and the ST and the C64 and whatever RS232 system you used).
To behonest, the idea behind Soundboards was already dead in 1994 when the first cards with DSP arrived. At first those cards did a hell of a workaround to be MIDI-compatible, hiding the RAM and the DSP behind a MIDI-Interface. But the writing was on the wall, in 1996 the first cards with DSP but without MIDI-hardware-layer showed up and with AC97 sound MIDI became totally irrelevant for PC-based audio. Also with the arrival of MMX the DSP was pretty fast dying off too because a 200Mhz MMX CPU could easily mix 100 audio channels at 100khz and 24 Bits.
AC97 is a jittery, clippy, low resolution nightmare though, sound awful to this day, as does using an older CPU as a DSP, noisy, high-latency, poor timings, a sound card and hardware DSP + DAC has always been vital for HQ audio. Unless you are just using low quality speakers and virtually no AMP power.
@@Wobble2007 You are treating in a field where you are guessing and not knowing.
While AC97 is limited to 16Bit, Stereo and 48kHz this is still even slightly above CD-Quality. As long as a good DAC is build into the system it is pretty much CD quality.
A DSP has ABSOLUTELY NOTHING to do with audio quality. It is just a linear processor for digital data. And MMX is doing EXACTLY the same and was basically developed to replace the expensive and driver-problematic DSPs back then. Funnily HDA nowadays has an minimalistic Mixer-DSP included again which e.g. Linux and ASIO drivers for real time critical audio simply doesn't use because it is slower than they CPU.
The Audio-Output even for pre-AC97 was already DMA-based for over ten years and therefore CYCLE-precise down to the Nano-Second. Even my Amiga 1000 from 1985 used DMA with a precision of around 150ns. That is 10000 times lower than anything a human could make out.
And latency only depends on the buffer size.
About the quality of AC97: It depends on the DAC. A cheap SiS-board usually had a crappy DAC, but at least the AC97 audio on my AMD Irongate board had studio quality (at least by 1999 standards) and a noise level lower than most CD players.
I remember that Tomshardware or Anandtech tested some AC97 solutions back in 2000 and they even beat some dedicated expensive sound cards.
@@CrassSpektakel I have yet to hear acceptable AC97, you will get missing data in CD audio playback on a AC97 CODEC IC, something as simple as guitar scratch will be auditorily missing, as will breath and other subtle tones, due to low resolution.
A DSP can mean everything as far as audio quality, it allows for EQ, reverb, flanging, attack, chorus, and so on, it can really help lower resolution audio, take the crystallizer on the X-Fi series for instance, it can work miracles.
Latency depends on a whole lot more than buffer size, drivers make a big difference for one, ASIO or WASAPI for one thing will always trump DirectSound.
DAC quality varies wildly, you will never find an onboard DAC that sounds good outside of Apple hardware, who at least use a reasonable quality default DAC, PC hardware always uses awful CODEC based IC's.
@@Wobble2007Everything you claim a DSP can do a CPU can do too. With MMX even faster. Even a PentiumMMX@166Mhz outperforms everything the back then Motorola 56k DSPs could do. Well, the DSPs needed less power for the same results but a modern Ryzen CPU can do complex mixing and converting of thousands (!!!) of channels in real time with less than 1% of its CPU power.
In fact in the 1990ths I wrote a software-audio-mixer for the Amiga which did all of this in real time on an 68000@7Mhz - 12 stereo voices in 14Bit audio at 56kHz. DSPs are just niche co-processors running DSP software. You can just write the same software for another plattform. It is not magic. Not even smart hardware by todays standards.
Back then I used the AC97 sound of my AMD Irongate as a software defined modem for amateur radio (AC97 was also known as MC97 - Modem Codec 97).
Why? Because my oscilloscope showed a timing precision of
In your video at position 8:05 you can see that the lower left pin of the serial interface is pushed inside. If I were you, I would repair that before it causes a short inside. With a bit of luck, you can carefully pull it out with a thin needle-nose plier.
I noticed that also.
Its like they tried to sample the mt-32 but messed up the sample loop point , make the samples sound popy and clickly,
Yeah, Sierra for sure didn’t spend much time on optimizing the instruments!
Yeh can be interesting looking at sound on a oscilloscope. Or sample the output and have a look at it on a screen.
An external MIDI device would have been really nice for other sorts of games. How many games had to choose between music or sound effects? MIDI for music and Paula for sound effects would have been the best of both.
With Paula you have freedom. No need to stack with fixed sounds.
Musos don't really like freedom. They like to quarantine notes on a stave and nail buskers to a tree.
MT-32 also was flexible, you could upload new patches and create new sounds. But yeah, the Paula was basically a simpler version of for example the GF1.
You also only had 4 channels. And a low rate. And it's sample-based synth equiv which means it has that weird sound to it when you try to stretch a sample over too much frequency range (octaves). I'd take the MT-32 over Paula every time.
@@TurboXray Indeed, but you could sample everything. That was I did as a kid and enjoyed it. For the price it was also awesome having a tracker.
@@104d_3rr0r_vince Ohh I know. I'm from that era. I made plenty of mods and cut my teeth on trackers as well. I also know all the limitations as well (and the tricks that try to hide those limitations).
If they had added an additional set of Paula chips that would have been enough, the Paula was very capable, what they needed was more channels.
I think 8 bit and 22-ish kHz was too little in the end. More channels would have been nice though.
@@root42 you need to take into account when the Amiga was released and when sound blasters became the norm, there is a big gap. The MT32 came out quite a bit later, and you could equally add an AKI sampler and blow it out the water, what made the Roland MT32 good was game compatibility. Monkey island is a good example of how the Amiga was better than your average PC at the time.
@@root42 and per-channel stereo panning instead of the hard stereo split!
the cripples in amiga sound and the gorgeus midi sound shock me.
It sounds like clipping, the sound cuts of at high amplitude, it also sound like its playing, low quality samples, without mixing. This games where made for 7Mhz 68000 nothing more powerful.
Yeah - those pops at the beginning of every note (in most examples)? That's about what happens if you don't start your samples at a zero-crossing - it's almost like they didn't really have a clue about how to use something sample-based like the Amiga, AT ALL (which doesn't make ONE bit of sense to me, as that should be one of the most basic things anybody worth their salt working with audio in any way learns pretty fast... I'm pretty much an absolute dilettante in that regard, but even I know that much, so that kind of error on absolute pros doesn't make one bit of sense 🤷♂...)
Very cool idea for a video. It always puzzles me, why you only have 7.000 subs. I also wonder why they did not sample the instruments from some professional synthesizer for the amigasnd.drv. Also interesting would be to see a side by side comparison on the graphics amiga vs. atari st vs. vga....
Its certainly an interesting comparison - but i cant help but feel they might have been a bit lazy with the conversion.
Some of the short throwaway jingles could have been a straight up sample of the equivalent MT32 ones.
Another piece which has a single pipe playing has a nice reverb on the MT32 which could have been achieved easily on the amiga by playing the sample notes across multiple channels so the reverb doesnt get cut off from the previous note when the next one is played.
Yeah none of these examples showcase the amiga's full audio potential! You only have to listen to an MT-32 playing the GODS or Speedball 2 music to hear how great samples can be. Also people loved using the inbuilt MT-32 reverb a lot and it hasn't aged well.
@@stefanimal5257 why do you think they were using the reverb so much lol.
@@NickFellows I think it sounded ‘fancy’ at the time. Popular music of the day had a similar love for reverb! Snare drums were always drenched in it.
different rather than better, the sampled sound was way more versatile but the synthesis higher fidelity
yea this is apples and oranges.. not better. different.. personally as a musician i'd take a sampler over an fm synth anytime.. th-cam.com/video/Pu0W2FDdMsM/w-d-xo.html image FM synth doing this song?
@@8bitwidgets Not FM, its linear arithmetic synthesis, which consists of a short ROM type sample attack and then a subtractive (think VA type in nature) phase after the initial attack portion. FM was actually on patent by Yamaha
@@jasonblazgk9973 ah LA synthesis.. i had a GR-50 that had that.. still very different tech and capability. paula lacks the built in synth features.. but i enjoy the 4 dedicated voices of it.. but in terms of music in games the richer synth capabilities of LA has it's place, but as a musician i still favor paula. i didn't play many games on the Amiga.
Nice fox candy t-shirt.
considering how old it is it sounds amazing i didn't think a computer from that year could sound so good seems like it was made in 1990 .seems like back then support for hardware addons was rare a lot of people had a dos pc but the sound on it is the most basic sound possible so this was a huge upgrade for pc owners. but on the amiga this would only be a minor upgrade but even other systems with more limited sound like the st and zx spectrum and so on didn't have any games that supported this or similar upgrades even on pc it took many years for it to get support from most games . but maybe it was just because it took up more memory to have different versions of each song
As you say, audio on the Amiga was affordable and good but not perfect. Like so many things the tight integration with the video timing made an hardware upgrade to Paula worryingly complicated. But then you could bypass a lot of limitations by using the CPU for mixing and clever combining of channels so all in all even an Amiga 500 was able to output 14Bit stereo at 56khz by pulling every trick available and it didn't even cost any additional CPU power. But that was just "one" channel. Sure, you could mix channels in software but that was where a stock 68000 was just out of its league, barely getting along with eight voices. With my 68030@25Mhz though I was able to get ~20 voices, although at high CPU load.
Die Auswahl an Mt32 spielen auf dem Amiga ist stark eingeschränkt. Mit MSDOS Emulation oder Classic Mac OS Emulation sind es noch ein paar Titel mehr.
very nice video, thanks ♥
sounds nice
Paula guarantees certain baseline audio PCM performance.
Excellent
Do you are anyone know about the studio mod for the MT-32? People would pay for a sound mod , I think a firmware mod/changing output ports, but am not sure. I did see old listings for it and people asking about it ,but it was unclear what the mod was , only that it had poor sound when it came to high-end hifi setups/studios, and you could fix this with the mod.
I haven’t watched yet, but as a fairly respected Amiga musician and also the composer behind a popular MT32 video, my hot take is that this is a flawed premise 😁
Na... Never! And again: I do love both the Paula and the LA32. They each have their own strengths and flavor.
@@root42turns out it *was* a flawed premise I’m afraid. It lacked in context and scope, with some inaccuracies and missed nuance. I’m not trying to be mean btw.
Großer Fan von MT-32 hier! :D Wenn du überlegst, daß der Roland D-50, also der LA-Synthese-Synthesizer, der quasi hinter dem MT32 steckt (dann natürlich als ausgewachsener Synth, MT32 ist halt abgespecktes Soundset) Ende der 80er 4000 (!) Mark gekostet hat wirds ein Schnäppchen ;). Die LA-Synthese ist eh clever gewesen, Mini-Sample meets substraktive Synthese, zwar digital aber abgeleitet von den analogen Synthies. Im Vergleich zu FM wie bei Yamahas OPLs (die im Yamaha DX7 als revolutionären Mega-Seller-Synthie, der den Pop-Sound der 80er definierte) um einiges einfacher zu verstehen. Wenn nicht Roland in Sachen Menü-Führung bis heute grausam wäre ;). Spätere Midi-Soundboxen von Yahama (XG Standard) find ich persönlich noch nicer (die Filter sind teilweise episch), aber da war der Amiga ja schon tot.
On a module like that it's more likely that you connect both midi in and outs to be able to dump sysex data to it. Seems like it does that when the game starts up. I can't really see any other use for the midi out otherwise :).
If you want to connect devices in a chain you'd use the midi thru port :)
What? No Monkey Island opening sequence? 🤪😱
Ha! If it would support MT-32 on the Amiga, yes, I would show it!
The MT-32 is a lot nicer on the ear here as the audio has a greater dynamic range/bit depth, the MT-32's DAC is likely a bit nicer than the amiga and the channel multiplexing is more sophisticated. Which I guess we should kind of expect from a box that was more expensive than the amiga500 and sold as a (not quite) pro synth. Though I understand there are bugs in the implementation if the MT-32's DAC. I guess the main drawback for the MT-32 is the fixed palette of sounds which the amiga doesn't have
The amiga is certainly much more versatile as it can be programmed to do all sorts; PCM playback, FM synthesis, even rudimentary wavetable synthesis (though I doubt the latter two were ever used in any games). The only thing that has always bugged me is the terrible channel multiplexing, having two channels that are panned hard left and two that are panned hard right isn't great and means the mono playback is mostly the only sensible option. You do see committed mod writers emulating stereo effects by using the same sample in the right and left channels but again not common in game music. I was always surprised that this, of all things, didn't get addressed in the shift from OCS to AGA.
I wonder how the amiga music examples in the vid was created? If they essentially just re-sampled sounds from other better synths/soundcards, even with the 8-bit limitation if they were somewhat lazy about it that could contribute to the lack of dynamics both in the instruments and the richness of the sound
Nice video! You should keep in mind though that you are comparing the bare Amiga to the MT-32 with Reverb enabled. This is not really a fair comparison imo.
I know -- this is a bit of an unfair comparison. The Sierra ports are also notoriously bad, compared to what was actually possible on the Amiga. However it shows what we often got, versus what money COULD buy at the time.
@@root42 You're right. I got locked in the Reverb thing too much when I heard the differences.
It has nothing to do with the Engine. Hero's Quest (Quest for Glory EGA version) with the SCI0 engine perfectly supports the MT32 on the PC because the driver is included.
I'm not sure how the Amiga "detects" that you want to output the audio using the MT-32 instead of "FM" or whatever the option would be for the Amiga.
Now I'm curious how it would work with the AtariST ...
On the ST those games had MT-32 drivers included, IIRC. But I think the early SCI0 games didn't have the full MIDI data on the Amiga. But I could be wrong. I didn't get any of the MT32 drivers from other games to work with earlier ones.
Ah, here is the SCI Wiki's take on the sound resources: sciwiki.sierrahelp.com/index.php/SCI0_Sound_Resource_Format
At the very least the Amiga games have the bank.001 resource, which contains the samples. One would need to check if the resource files also contain the patches for MT-32, or not. If they are missing this would not work, no matter if you had a working MT-32 driver or not. I need to check that with some SCI utilities. It would be nice to activate more SCI0 games to play MT-32 music, though!
I would stick with the original as it gives the Amiga a unique sound. I wouldn't want games to sound like a PC.
I can see that!
I think can be solved with bit more CPU power, and better wave generator, at least on an accelerated amiga.
The amount of copium in this comment section is popcorn-worthy. Great video! :)
I didn't know Amiga could do MIDI, I though was more of an Atari ST thing.
MIDI is a different-shaped serial port.
Sounds like they recorded samples off the mt32 and stored them as PCM samples for the Amiga to play. The mt32 is just an FM synth, which the Amiga doesn't have. It only does 4 channel of PCM audio, which I would argue is better, aside from this particular case where Sierra just did a rather nasty port.
The MT32 isn’t an FM synth.
@@magicalsynthadventure3216yes, you're right. Certainly sounds like one though. It uses PCM samples and subtractive syntheses, apparently.
the amiga sounded amazing already, better than the crap with pc's.
"Better" is in the eye (and ears) of the beholder. The amount of equalizers and DSP's all around should be proof of that. If i push Turrican II through WinUAE to my headphones, the output is crisp and... nerve-wrackingly so. But if i push it though my Yamaha YST-M15 plus YST-MSW10, ehhhh, almost sounds the same as the Amiga did, which is how i liked it. Similar thoughts would be me missing WinAMP's Enchancer DSP because it was the only thing that made it "sound right", and cheap Sony headphones because they "sounded right" out of the box.
Different yes, better depends...
I'm a USian, but I've always *far* preferred the game and demo music composed and produced by the Europeans on the non-PC computer platforms compared to that of the USians for any platform, including PC and MT-32.
Edit: As a USian, I include British & Irish as Europeans, ofc. I say this b/c I know that at least some Brits (and perhaps a small number of Irish) don't identify as European.
Little better midi player than MT32 was ProjectXG, it was based on Yamaha DB50XG daughterboard (SW60XG was standalone card, CS1x standalone synth). XG standard is far more powerful than General MIDI or Roland GS, good example is XG midi music from Final Fantasy VII. But i love Amiga's Paula as is, just 4 channel pcm sample player without real synthesis, trackers which can do synthsouds are probably precalculated one cycle waveforms, varied with some realtime commands.
After all these years I must say, the MIDI-sound system does not sound as good as I remember, back then it was spendid, when I listen to it now it is a bit shallow, plastic and cheap. When MP3 came to computing music the MIDI system lost its spender, vividness and excitement. The Amiga sound chip could play back samples and back then it was quite an achievement, now only real sounds sound real.
All I can think while watching this is, that guy doesn’t have a Finnish or Swedish accent, so why is he wearing that shirt?
I have connections…
I wonder what one of these MT-32 Raspberry Pi emulators would sound like? Something obtainable
They sound exactly like the original. At least to my ears. I think I have a video on that topic as well.
@@danielktdoranie they sound absolutely identical to the real thing there are examples on TH-cam which do the comparison
So, why do you have a t-shirt with a logo of a Finnish candy on it?
The question is: why doesn’t everyone???
@@root42 Pihlajanmarja Kettukarkki ftw!
32 bit sampled instruments vs 16 bit sampled instruments?
I wonder why the samples are clicking so much. Is it because samples do not start from zero crossing or something else? I've never owned Amiga so I don't know how common this is.
@dr.benway1892 Yes, that's roughly how most of the examples in this video sound like - just about the worst I've ever heard from an Amiga... But as I learned from some comments around here, Sierra was _distributing_ the MT-32 at the time, so they may have had kind of a vested interest in not making the 'competition' sound 'good' in any way, shape or form, so 🤷♂️😁...
ok to be honest, if you want to hear the real capabilities of the Paula chipset you should play amiga games and not dos games created with adventure engine that are interpreted by the engine itself and not optimized for the host machine.. do not misunderstand me, i love dos game and also lucas and sierra games but their games were create mainly for pc dos computer that has no sound capabilites at all (until sound blaster arrive) and so when they were ported to the amiga they were ported "as is" without any kind of optimisation... it is nice to have midi on the amiga (bars and pipes and many others) but i do prefer the amiga sound...
Of course. Or just taking any of the great demo productions shows the power of the Paula. This is more of an example how little love some of the Amiga ports got.
WHAT IF SIERRA PROGRAMMED MUSIC & GRAPHX BETTER ?
To be honest, MT-32 was very expensive and it sounded somewhat flat anyway. Way better music got made on native Amiga sound hardware by musicians like Chris Huelsbeck, Jeroen Tel or the Psygnosis people using samples, or on the Megadrive's Yamaha chip by musicians like Yuzo Koshiro. Sierra's artists were just lazy asuming that everybody had expensive hardware, then grossly underutilizing "standard" capabilities such as EGA and Soundblaster, just because they didn't match their development set-ups. As a counter-example look at what Rare did for Nintendo on the SNES with Donkey Kong Country, which looks and sounds awesome on such underpowered hardware (which did sound mainly by sampling, just like Paula).
I've heard much better sound from the Amiga in other games. LucasArts games especially sounded great. Sierra games on Amiga were known for sound that was lacking.
How do you even dare to name this video like this? 😉
Amiga lovers are so easily riled up... :-D But I love my Amigas. They are awesome machines!
Paula chip frequencies are a bit tricky to measure. Isn't it so that if you want a simple square wave you already need 2 samples and if these are in 28Khz the actual frequency would be closer to 14Khz. But yeah, don't get me wrong, despite a bit weak bass tones and saturated mid-tones & limited hi frequencies, Paula gives the "Amiga" sound. And I hear people complaining about "having only 4 channels"... Get on with it! If you can't put up a good song with three sound channels, study first with only two!
There are awesome tunes using the TED (two channels), TIA (two channels, weird notes) and even the original Speccy and PET (one channel square wave). Indeed much of the creativity stems from those limitations!
@@root42 Just recently attended a seminaire about Speccy's sound chip and the evolution of the model (at Skrolli-party). Amazing improvements from fairly horrible one-channel beeper to CPU-hog 2-channel mode and towards newer 3 channel chip and further into filter-like "synth-trickery". BTW, Cool video and sweet shirt :D
The only really big flaw with Amiga's audio was the hard panning. 3 channels for music and 1 for sound effects would have been sufficient but sound effects only coming out of one speaker sounded awful. And balancing three channel music with two on one channel and one on the other was painful.
You were best off just externally mixing the two channels, but most games didn't design for that and instead you got this norm where you got music or sound effects, often you could choose which but rarely did you get them at the same time.
Some people actually like the hard panning. But yeah I think it would have been clever to allow for soft panning.
I listened to music of unreal games and games like crusader lately or even the music of the unreal demo from 1992. I also like the music of pinball games on the amiga and pc. I prefer listening to music of impossible mission or shadow of the beast instead of midi music of pc games. It all would't exist if an amiga 'pushed' developers to create midi files without sample content. I think the history is just fine 😇 not to be negative in any way. I love what you did in this video 👍 the latest music I made is actually a general midi file for a doom mod 😂
Pinball games in the early 90s rocked. Especially the music.
Well, even when used with the most expensive 14" CRT TV 2 inch mono speakers of Sony/Panasonic etc you won't be able to tell the difference between 28khz 8bit Amiga vs 44.1khz 16bit CD quality audio of a console so there's that......
I think this is no improvement. MT32 was better than Adlib for pc. But listen to Amiga Version of Shadow of the Beast ost.
Wonder how big of the chip or chips, would Amiga need to have that kind of the sound?
IN addition to what was Paula providing.
Amiga did not play synthetic sound, it played prerecorded sound, it technically can have been generated but 68000 7mhz was not powerful, it takes lot time to do so, it be lot larger technical challenge to convert soft sounds into wave sounds. Then having the sound prerecorded, my guess is that sampled sound low frequency, to save space.
@@kjetilhvalstrand1009 I know that, that is why I asked for hipotetical another synth chip
@@DS-pk4eh Sound is rage of 20 hertz to around 20 000 khz, cd quality is 44.1 Khz, now that’s far away from 7mhz, so should have problem doing it, but it will take a substantial, amount of CPU resources. This resources you want to use on the game.
@@kjetilhvalstrand1009 That is not how that works. The 68000Mhz has nothing to do if it is capable to produce the sounds human can hear.
Anyway, I was thinking how big chip like that from Roland would be to have it next to others chip in Amiga. So, unless you know the size of the chip in Roland and how it worked, it will not answer my question.
Its Sierra problem. Conversion of MIDI music to AMIGA 4ch format, was make terrible music.
Sierra didn't bother to adapt these games to PAL speed, they sound slow, compare them with PC.
I'm just here for the comments. 😂😂😂
I really like the Roland sound but could one not just sample it and use the same on the Paula or am I thinking too simple??
Sierra simply did not do a great job in porting to the Amiga. It was just a "good enough" solution I think.
No. You are correct. There's nothing stopping someone from just sampling the MT-32 using the Amiga/Paula chip.
I never understood the MT32; it made everything sound the same. only later, with wavetable and XG this became mature. I never really liked the sound of the Roland stuff... except for the TRx0x and TB303 of course ;-)
It sounds like soundblaster vs gravis ultrasound.
The real question is: did Gravis Ultrasound sound better or comparable to MT32 ;) A500 was not in the same league, but well, GUS was probably close.
GUS series cards don't have reverb, they can sound pretty flat and boring, so you have to add some reverb during sample edit. Advantage is, you can use any sample.
If the Amiga had come out with better sound, it would still have sounded better than its price competitors assuming same price.
But really this would have cost far more and so people wouldn't have bought it. They'd have bought a sampler and Atari ST instead for the same money. If they had wanted to make CD quality audio.
And this is why the Amiga didn't come with better sound than the Amiga. Ever. You can do a lot with software for free, hardware costs $$$.
Sound that only someone upgrading from a farty pc beeper could love
Boar, die Amiga-Portierungen spielen gesampelte midi Samples... Und das ist mal schlecht konvertiert!
Ich finde, das kann man nicht vergleichen! Ein mies digitalisierter midi-Sample klingt neben einer General midi Box natuerlich schlecht!
Hier sollten Scene Musikstuecke verglichen werden, nicht schlechte Audio portierungen.
Hey Why are you having Fazer Pihlaja shirt ? cheers from Finnland
Fazer make great stuff! :)
@@root42 yes it does but we are missing äppfelschorle and spetzi 😂
You don't happen to have a sister who calls herself Survival Lilly do you?
Luckily, no.
This sounds very bad, like a cheap Casio, and the music (as in the notes) is also very poor and stale. I'm trying to bend my mind around the concept that someone could actually think that simple-minded barren songs using terrible instruments sounds better than all the bangers in hundreds of Amiga games, and thousands of demos. The Amiga sound, and not the cheap sound canvas sound, is largely what sold millions of Amigas.
Come on it's a midi squark box.. if you can play it on a mt32. You can play it on a dx7 or a Dr midi.. I used to do a lot of midi on the miggu.
Amiga sounds much better, Sierra doesn't squeeze Amiga audio
do not compare it to tracker music, you cant use it much in games. You are very limited by outdated Amiga architecture.
@@madigorfkgoogle9349 The first Amiga tracker was designed by Karsten Obarski for making in-game music. It was cloned to make other trackers like NoiseTracker and Protracker. Whether writing tracker music for demos or games, it's the same thing. Note: The Amiga can sum channels to play back 14-bit stereo samples. Not bad for a 1985 machine and puts it ahead of the Ensoniq Mirage and almost to the level of the original EPS (minus the polyphony).
this is just wrong. The MT-32 is sounds awful and is very limited.
The game conversion team at at Sierra simply did not in anyway have the skill set to leverage the strengths and unique ability's of the Amiga's hardware those that did knew exactly how capable the Amiga,s sound hardware was capable of. th-cam.com/video/BB3_33HFyus/w-d-xo.html
Używam amigi od lat 90 a nie wiedziałem że na miga pisano gry które potrafiły wykorzystać port Midi z obsługą MT 32 w Polsce się o tym nigdy nie mówiło Przynajmniej ja nigdy nie słyszałem a kupowałem co miesiąc miesięczniki Amiga i inne Commodore&Amiga
MT32
when you look at the poor quality of the games it doesnt need any better sound to be honest , its a shame it didnt get any 3d chips they could have put a better chip in like an s3 or something with more video ram that would have been great and pc games could have been ported on it i know people have done pc ports to amiga on aga and there pretty good but it would have been great for faster 3d then they could have had 16bit sound then it would have been awesome. thats why soon as the PlayStation 1 came out booom end of the amiga for me.
Better sound than an Amiga? 😮
That is heresy, plain and simple!!😅
Okay, I might have polarized here just a LITTLE bit. :-D No bad feelings -- I love the Paula! And also the LA32.
Both sound like toy keyboards. The Amiga can do much better than this. I don't know if the MT-32 can do better.
Agreed. The Paula can do better. But I think back then the know how wasn’t quite there yet. And yes, the MT32 is a consumer grade synth. It obviously sounds… synthetic. But back in 1989 Sierra games sounded better on the MT-32 than on the Paula, I think. Today demoscene prods show what the Paula is actually capable of.
I agree, this memory / storage compromised sound quality.
@@root42 not really, Paula was a very outdated chip very fast. It was made with cartridge based game console in mind, so if you want to use in game music you have to resort to 12KHz or even 6KHz samples, also Paula has zero mixing control. The demoscene is usually using all the power of computer, using CPU for downmixing, you cant have this in games, especially not with higher sample rate. The ChipRAM is also a very limiting factor, since games had to run on lowest denominator ant that was 512KB ChipRAM Agnus. You cant fit much in that, on cartridge based console this would be no problem since you would fast switch those samples, but Amiga was different at the end...
Also dont forget that these games were made in early years and many tricks didnt exist back then...
@@root42 I think it's also a matter of where you come from. In 1989 I was just getting into demos and was amazed by the music disks from Mahoney & Kaktus. The resolution-limited (grainy) sound actually worked great for me. When I first heard midi tunes they were very underwhelming as they didn't sound much better than those early Soundtracker modules and the demo musicians had vastly improved. Even some far more limited (technically) C64 SID tunes sound better to me. It's all in the mastery of the musician. If you listen to 'Guitar slinger' from 1993 you might mistake it for a 'real' tune on FM radio or an mp3.
The Paula is a good chip, but at the end of the day it's just 4 channels that cannot be panned.
worse Amiga sound in those games are not Amiga Paula fault. Developers used shity midi player with shity sounds. Roland got build in better sounds...