... a retrospective view on sound card history
Dolby Digital, dts, EAX... all being terms, that advertising world throws around.
Sound possibilities, that were only located in cinemas some time ago, already find move in our home. You hear the Enterprise already coming from the back. You can localize your opponents in Games after your hearing. Full sound always and everywhere.
You could think, it has never been different, probably you don't want to imagine that it has ever been different. We actually are much too pampered - are we?! After all, you should consider that meanwhile our oh-so-styled miracle crates were gray as mice not far ago, and their acoustic possibilities haven't been a large difference.
Limited to a simple whistle, at least in different pitches...
but stop here! It's very unlikely that we just switched from PC Speaker to 5.1 sound from just one day to another - in the opposite!
How did we actually get to the point, where we are these days? What do all these sound options mean in the old games? Does MIDI really sound so bad as you remembered?
This article should bring little light behind these questions, it explains the technology in the interesting years of development from its origin until approximately 1995, and should show, what we actually already knew, but often won't see: that a name and good marketing often mean more than advance and innovation...
For the complete correctness of the content, I don't take over any liability - the article originated in extensive investigations. Heavy effort flowed into avoiding wrong statements, however this definitely could not exclude their appareance either...
Any information regarding this is welcome!
Continuation
How did it start?
This question cannot be determined unequivocally. If we ask about "sound", do we mean the same-named Basic command, or - more possibly - what we understand as "good sound"?
In this case we probably have to start with the AdLib-card.
But in order to have clarity about why this card came on market anyway, we'll better start completely at the beginning: with the PC Speaker.
How we recorded
LAPC, AWE and GUS happily together
The data of the PC Speaker was passed over the detour of an AWE32 to the Live while both OPL-Playback as well as the MIDI-Playback of the AWE32 were sent digitally to the Live and were equalized after it (more on this trick on the AWE32 page).
The sampling ran with 44,1 Khz, which was conversed from 48 Khz (namely only with these the Live works internally) in favor of the better compressability for the Web.
Apart that, the human hearing is not able anyway of dissolving more than 20 Khz. Especially confirmed disco visitors should find their border clearly farther below...
The Roland SCC-1, that I don't possess unfortunately, was recorded by Eike Belgardt. Thanks for this!
The IBM feature Card - and the Tandy/PCJr - samples were kindly provided by Jim Leonard (Trixter/Oldskool.org). My particular thanks flies to him, because I exactly only could pronounce the word "computer" in contrast to him at the appearance date of the PCJr...
|
Playback
The samples are offered in the formats OGG and MP3. Newer versions of Winamp (approximately 2.80+) are also in the position of playing OGG, otherwise the fitting plugin can be downloaded at the specified link below.
I recommend the playback in this format, since it sounds generally better - especially with strong compression - that we used here - in opposite to MP3, and is able to reproduce the actual sound essentially better.
MP3 serves solely as expedient here, especially since often with 22 Khz, if OGG cannot be used for some kind of reason.
And don't forget to activate the file format also in Winamp, otherwise playing it simply won't work ;).
A small hint when reading: the history of the sound cards is more or less written in one piece. Whoever would like to read it this way may jump over the samples and technical details.
A summary of the samples for direct comparison is moreover also on the last page, before the glossary.
The beginning...
PC Speaker (1981)
It was in the year of 1981, as the PC in form of an IBM PC model 150 with a 8088 CPU saw the light of the world - a fact known by almost everybody.
If something counted, then this: it had to be cheap! Other machines probably also didn't succeed as a result of their pretty high price tag. So then it even happened that the at that time better election, a Motorola-CPU, was discarded. It was simply too good for this product!
It was therefore a cut almost everywhere. The continuation of it namely found itself in many other fabricated parts of the PC, over which is partly sworn still today. So it was just about the same with the audio playback. As office machines don't have to be able to return whole symphonies detail-trusted, sound was restricted to the absolute minimum: the PC Speaker.
IBM PC Model 5150 - (c) www.old-computers.com
One would like to think that it would sound hideous to seriously play music through this thing, and admittedly, it mainly did. Only the human creativity often arises out of the pure limitation, and actually it was hardly different here.
While some games made use of quite simple tricks to get a better sound, for example playing in short frequency at different pitches, in order to simulate more channels, there were also other attempts, like with Lucasgames. To lend their songs some more pep, they tried to simulate drums (sample "Maniac Mansion").
However, the end of the flagpole was not yet gained. Meanwhile, resourceful programmers had found out namely how one could elicit even real samples from the speaker! Admittedly not completely without expenditure, because this methodology necessitated a meticulous timing what made running a game parallel very difficult on slow machines. Playback therefore was initially often restricted to the intros of the games.
A popular example, which existed alone to the demonstration of this technology, was "Magic Mushroom" - a small, ancient commercial, which was played through the PC-Speaker.
Inevitably each game supported the Speaker practically since a "real" sound card was still very expensive in these times.
Samples
| Monkey Island 1 | | [OGG] [MP3] | | A usual sample the user was listening to | | Maniac Mansion | | [OGG] [MP3] | | Drum simulation | | Magic Mushroom | | [OGG] [MP3] | | Original sample | | Magic Mushroom | | [OGG] [MP3] | | ... and how it sounded through the speaker |
|
Technical
To do justice about the question "how can I play samples anyway", here a small technical explanation:
The Speaker still is a piece of mechanics. This makes it inert, so only sounds until a certain frequency can pass. And exactly this is what we take advantage from. Instead of feeding it with its usual frequency spectrum, we use it with an essentially higher frequency, and switch it short in this frequency on or off, to get a resolution of 6 bits after all.
However, a problem was the often very low playback volume. Another interfering factor was the whistle of the loudspeaker on the high modulation frequency.
Both problems have been resolved later for example with dynamics compression (quiet sections louder), and also better programming, and so only one problem remained: many cheaper PC clones had installed a piezo-beeper, and this one could return higher frequencies decidedly better - the inertia of the normal speaker was missing, and a useful playback was almost impossible.
|
PC Junior / Tandy 1000 (1983)
IBM PCJr - (c) www.old-computers.com
Anyway, main problem of this machine was the keyboard which ran via infrared, instead of the promised several meters only the quite near sitting in front was possible, and so so it was practically free of sense.
But the best nevertheless was the sound at that time: using a chip by Texas Instruments, it was possible for the machine to pass 3 sound channels and one noise channel with 16 volumes each! Looking at these specs today, this doesn't sound much better than the capabilities of the first Gameboy, but for the PC-world, it was a real enrichment!
Meanwhile, also Tandy developed a clone, which they wanted to praise as "PCJr compatible". However IBM withdrew its model at this time, because although the second version admittedly had no more the deficiencies of the first one, only few were sold because of the bad expandability.
So Tandy only titled its model "MS-DOS compatible. In contrast to the PCJr however, this model could spread quite far, and so the music playback is supported by not only few games on these systems.
A later expansion in the Tandy TL/SL made moreover even the direct playback of digitized samples possible with simultaneous utilization of the synthesizer!
Covox Speech Thing / Disney Sound Source (1986)
An idea, that is not even far fetched, represents this possibility of the sound playback.
The thought is quite simple: the parallel port is nothing more than a possibility to pass out 8 information bits simultaneously. Each PC had this port, and the conversion of this kind of data to electric oscillations represented only a small, very uncomplicated step. Hence the only thing missinge was the interface, which a company named Covox delivered with their "Speech Thing". For $70.
However, this utensil was nothing further than soldering together some different resistors, the simplest form of a DAC at all, and actually was not worth its money. However, the simplicity admitted the reproduction independently (see link).
Some other versions even supported the recording (!), and there are even circuits, which made stereo playback possible using a switching circuit with two DAC's.
A quite similar product was expelled also by Disney, named "Disney Sound Source". However this one was built technically somewhat more complex in order to make also the parallel usage of printer port possible. Using a 9V battery, it featured an integrated amplifier!
There are only few games, that supported the parallel port as sound possibility - whether now using Covox or Disney Sound Source. However, the Covox playback was later quite wide spreaded in the demo scene.
1987 - The real sound...
AdLib
AdLib in the version of 1990
So it also happened that some suppliers offered plug-in cards to expand the sound possibilities. While some were practically unsuccessful with it (for example SSI 2001 with SID 6581, which came from the C64 (!), or Covox with their Sound Master) a company had of the success clearly more: AdLib. By the way, the company name "Ad Lib" was derived from the latin, and is an abbreviation for "ad libitum" - "at will".
Since other systems sounded more or less like the multiplicity of interconnected PC-Speakers, AdLib used the Yamaha YM3812, today commonly known as OPL2. This chip, originally used in low-end-keyboards, does in different modes 6 voices (+5 hit instruments) as well as 9 voices. Above all however, the user could influence the sound of his instrument through a multiplicity of possibilities himself. While this actually was intended to reproduce real instruments, they created a quite individual sound, that really could sometimes convince - depending on creativity of the user - as autonomous.
The reverse of the medal on the other hand was that it cost also some expenditure to let instruments sound really well on this card - the mass of the games nevertheless used the abilities of this card only very superficial, so the sound often came across very "cheesy".
Without question, however, AdLib started a mass market with this card, and also practically all newer DOS-based games still supported it. However unfortunately, AdLib thought much too late about the possibility playing back digital samples, and not little later, that should happened to be their fate...
Samples
| Monkey Island 1 | | [OGG] [MP3] | | One of the better AdLib tunes | | Space Quest 3 | | [OGG] [MP3] | | Rather average | | Silpheed | | [OGG] [MP3] | | Pretty good usage | | Easy AdLib Track 1 | | [OGG] [MP3] | | Here the AdLib presents its very own style | | Tempest 2000 | | [OGG] [MP3] | | Techno on AdLib, rather seldom style | | Descent | | [OGG] [MP3] | | Descent I intro - badly done for AdLib |
|
Technical
Based on the OPL2 the card could, by using frequency modulation (FM), either play 9 sound channels or 6 sound channels and 5 hit instruments (bass Drum, Hihat, Tom Tom, Snare Drum, Top Cymbal) simultaneously. Task of this chip was to reproduce real instruments. However, on the basis of only 2 frequency generators this was quite impossible.
Additionaly, one could put the gotten frequency into a quite simple cover in order to imitate the dynamics of a real instrument (Attack/Sustain/Decay/Release).
It is also interesting to know that later some software (specifically tools for the MOD-Wiedergabe) but also some games, applied quite similar tricks like with the PC Speaker, and so actually originated a qualitative relatively appealing sample playback.
|
IBM Music Feature Card
IBM Music Feature Card - thx to Jim Leonard
It was essentially a Yamaha FB-01 on a card (much how the Roland LAPC-1 is an MT-32 module on a card) and inherited all of the FB-01's properties: 8 FM voices, stereo-pannable, and each voice was controlled by four operators instead of Adlib's two. It had an extensive built-in library of over 300 high-quality synthesized instruments ("classic Yamaha" as many have called it). You could put two of them in a single machine and get 16 voices. Because of the stereo and 4-operator FM synthesis, you would be tempted to call it an Adlib on steroids, but it was more a MIDI module than a PC-based sound card. But even in that area it was advanced: One thing that seperated it from other MIDI-based instruments at the time was that you could detune each note in very fine increments, so at one point it gained a small devoted following of atonal composers.
Sierra was easily the company that supported it the most. This is not surprising, because Sierra's entire existence was saved by IBM back in 1983 when they contracted Sierra to make the first "next generation" game for the upcoming "home" personal computer, the PCjr. In addition to 16-color graphics, the PCjr supported 3 voices plus a noise channel instead of the typical single-voice beeper. Sierra had always been dedicated to IBM after the success of King's Quest, so they likewise were one of the first to support the IMFC and also more than 4 colors on MCGA machines (PS/2 model 25 and 30, which could display 256 colors but were NOT VGA compatible and could not run EGA games).
So why didn't the IBM Music Feature Card take off? It retailed for $600, which didn't help. It was geared toward professional musicians much more than gamers (Sierra and Microprose are two of the only three companies I've seen support it). And because it didn't have built-in
reverb like the MT-32, many people thought it sounded "worse" than an MT-32 (of course, if you apply the same reverb-like effect to some of these IMFC samples like "minuet" you can clearly hear it was a strong match for the MT-32).
(Information and samples kindly provided by Jim Leonard)
Samples
| Silpheed | | [OGG] [MP3] | | One of the better Tunes by Sierra | | Space Quest 3 | | [OGG] [MP3] | | Example how bad a support can be | | Colonel's Bequest | | [OGG] [MP3] | | Show how well the card performed | | Minuet | | [OGG] [MP3] | | Also a very good example for that card | | Larry 3 | | [OGG] [MP3] | | Intro tune; pretty good done |
|
1988 - Game Blaster and LAPC-1
Game Blaster/Creative Music System
A certain form of stillbirth was brought on the market by Creative Music Systems (now Creative Labs) in the year of 1988.
Using the same name, they brought a sound card on the market, that a little later was marketed by Radio Shack, an electronics shop chain, as "Game Blaster".
Take 12 PC speakers, allow them a volume alteration of 16 steps, add a noise generator, that should reproduce 3 different drums, and allow moreover the playback in stereo. After all something, with which they could shine in opposite to AdLib!
AdLib offered essentially more for the money despite all, and so it only is comprehensible that this card didn't find any larger spread. However already at that time, Creative Labs understood secure themselves the support of game companies. In effect you still met a good mass of games, that nevertheless played "music" on this card. Unfortunately anyway, most games really only used it as multiplication of the speaker, instead of making use of its advanced capabilities.
Samples
| Monkey Island 1 | | [OGG] [MP3] | | Pretty good - better than PCJr, but can't reach AdLib | | Silpheed | | [OGG] [MP3] | | Not overwhelming | | Space Quest 3 | | [OGG] [MP3] | | Pretty bad example |
|
Links
| FAQSys | Programmig the Game Blaster | | Mobygames | Games supporting the Game Blaster |
|
Roland LAPC-1
Originally only marketed as external module under the name "MT-32", Roland dared the jump to the PC card, and reached something, which presumably nobody would have expected...
As well as in bundle with sierras games, as also individually available, the card admittedly came without its own software, but nevertheless offered something: 32 MIDI channels, however non-simultaneous 32 instruments - see "technical", with 12 bits D/A and 32 Khz mixing and reverb effect. 128 pre-defined instruments, above all however the possibility, to load his own instruments (patches), made this card a multi talent. If something should once be missing, you simply reload it!
Probably often, this was the case with the piano, because this instrument was reproduced by the card rather badly. However, it owns a brilliant drumset - 30 drums are pre-defined!
While some manufacturers merely converted their tracks from AdLib on Roland, the disgrace of the bad exploitation has not been so strongly represented in contrast to AdLib. In games, the music playback almost always sounds full and balanced using this card, depending on the skills of the composer, even today, this card should elicit some people a rapturous "wow!". It should surprise some so - the old games actually sounded THAT good!
However, all this had its price, namely approximately $500, that not everyone wanted to or could pay. So, this card occured only a low spread. And even today, it is rather rarely to get over Ebay - and if, then not quite cheap.
Admittedly the MIDI playback is (because it is nothing else that it does), after present-day standards rather mediocre. However one also should see on that occasion, that this is merely the only card fully compatible with the "Roland" option of many old games, only this one supports, what newer cards, that call itself "LAPC-compatible", even Roland's own, can't do - namely the upload of instruments. Otherwise it can occur that a laser shot suddenly sounds like a piano.
Only Roland's external synths represented an alternative, namely MT-32, CM-32L, CM-64, MT-100 and CM-500 - in connection with an MPU-401 compatible sound card. An interesting detail was that lots of game producers showed their names on the module display.
Whoever puts value on compatibility with old games simply can't avoid this card, or one of these external modules!
Samples
| Monkey Island 1 | | [OGG] [MP3] | | This is how it should actually sound like | | Monkey Island 2 | | [OGG] [MP3] | | Even a bit better and uses own samples | | Silpheed | | [OGG] [MP3] | | Also a very good tune | | Space Quest 3 | | [OGG] [MP3] | | The same goes here... | | Ski or Die | | [OGG] [MP3] | | Electrifying saxophone - unusual, bot good! | | X-Wing | | [OGG] [MP3] | | Probably Star Wars was composed using this card?! | | Descent | | [OGG] [MP3] | | Descent I intro by using official Roland GM Patch |
|
Technical
The LAPC was based on the MT-32 chipset, an external synth of Roland. "LA" stands on this case for "Linear Algorythmic" and explains itself as follows:
Wavetable, as we know it, usually plays only one sample from memory at different pitches.
Possibly, an instrument could consist of several "layers" - a superimposition of different samples for an instrument.
LA tried to refine this principle, specifically Attack, Sustain, Decay and Release are individually calculated and overlaid. So, the different playback speeds didn't come too much inforeground, and one got a very true image of the original instrument.
If then layering was used as well, up to 4 (!) channels were covered for just one instrument. The LAPC-1 therefore was theoretical only in the position of playing 8 instruments simultaneously.
|
Links
| MT FAQ | FAQ by Juho Sippola concerning second hand Roland Synths | | Mobygames | Games supporting the LAPC-1 |
|
1989,1991 - The first Sound Blasters
Sound Blaster (1989)
Sound Blaster 1.5 with CMS chips
So they went to the game industry, asked here and asked there. And got the unambiguous result that the card besides playing music should be able to play also samples. The result of this was their Sound Blaster.
Equipped in version 1.0 with a DSP that returned 22 Khz 8 bits Mono and recorded with 13 Khz, CMS chips for compatibility with their own Game Blaster and above all with a Yamaha OPL2 they entered into direct competition for the AdLib. The latter was supposedly possible only thanks to Microsoft, they persuaded Yamaha to bring the OPL2 on the free market, until then, AdLib was exclusive partner and buyer!
Production in Singapore made the card cheaper than the AdLib, which was produced in Canada. Still, CL understood very well to offer support for the programming of their card. AdLib was pushed fast from the market.
SB 2.01 with empty CMS sockets
Under the "bonnet", it looked very chaotic. Seemingly, they were in trouble getting the card fast to the market. This resulted in a difficult programmable DSP (although one product of this house should later top this), and a sound rendering, which often made itself noticeable by crackling. But after all - they had brought the first consumer card on market that allowed playback of digital samples!
By the way, the crackling was created for specific reason: data was sent to the card in blocks via DMA (Direct Memory Access). If one block reached its end, you had to inform the card, that another block will come, and simultaneously, you had to suggest this situation also again to the DMA controller. This led to considerable gaps, in which simply nothing was played, especially on slower systems. Stopping sound data and its further playing led to partly considerable level fluctuations, which we then perceive as crackling.
Another problem, which often was not heeded by developers, and which existed until the Sound Blaster Pro, was an only very coarse alteration option of the playback frequency. Placing 22,05 Khz precisely was not possible at all. Many developers overlooked this, and so music often is returned too slowly. , Example: Epic Pinball - SB Pro compared with Ultrasound.
Further versions of the card little later followed. Version 1.5 was a economy version: the CMS chips for Game Blaster compability were optional, instead there were two empty sockets.
Version 2.0 on the other hand saved another chip, that was also intended for CMS (a standard chip, but programmed by CL). Simultaneously, they optimized the circuit board layout, which now worked with clearly less chips. Most importantly, however, was the DMA autoinit mode that finally enabled the card to play continuously, therefore without interruption and crackling! And finally, with version 2.01, they made sound playback with 44,1 Khz sampling rates possible. For this reason they introduced the "Hispeed Mode". Translated this means, that the DSP was so heavily loaded in this mode (used of above 23 Khz), that it didn't accept any commands from outside anymore, and the rendering could merely be stopped with a Reset of the complete DSP.
All these improvements brought some alterations to the interface of the DSP - one of the reasons, why other companies had so heavy problems really being 100 percent compatible to this card. While angry tongues claim, this is above all a behavior in order to stop competitors, I claim that this is also a result of incompetence on one hand, but also time pressure on the developers side at Creative Labs on the other hand.
Sound Blaster Pro (1991)
A further development of the original Sound Blaster was the SB pro. It differed above all to the SB 2.01 in the ability to return and record stereo. The latter now even at 22,05 Khz!
They left the DSP at 8 bits, however added another OPL2 on it: a second chip of this kind guaranteed stereo music playback (although this was seldom used by games).
Also this card went through several revisions, very likely 3 of them, while every revision brought improvements to the interference factor concerning bus transfers (one could follow his mouse movements, keyboard entries and all other live at the boxes).
Revision 2, with which the two OPL2 were replaced with an OPL3, made the biggest difference. OPL3 was downward compatible with OPL2, played stereo with 20 voices and made the synthesizing possible with after all 4 operators instead of 2 as with the OPL2.
Also with this card, Creative Labs obviously was a little hasty. Maybe some remember the "reverse stereo" option of some games.
According to unconfirmed statements plain and simple the first two revisions of the card swallowed (maybe also all, nobody exactly knows it) the first data byte, therefore the left channel, and then happily started playing on the left side - but with the second byte, hence the right channel!
They recognized this defect and introduced an obscure code snippet in their program examples with the headline "enable stereo playback"...
Samples
| Epic Pinball | | [OGG] [MP3] | | MOD-Playback, here the game interpolates by software | | Descent | | [OGG] [MP3] | | Descent I intro - OPL3 sounds like OPL2 in stereo... |
|
1991 - Competition!
(Advanced) Gravis Ultrasound
Fire red Gravis Ultrasound with 256K
The card was somehow revolutionary. Based on the Gravis GF1 (actually only a clone of the ICS 11614) and equipped with 256 KB RAM (maximum 1 MB) this card could mix at most 32 voices with 16 bits and at best 44,1 Khz. Therefore it followed the concept of wavetable cards.
Actually the card was directed at amateur musicians, however its abilities alerted a quite different group: the then demo scene!
This card was prized only little higher than SB Pro, however, offered an essentially better audio quality. But main argument was in the mixing in hardware. Until then, this was a task fullfilled by the CPU, and in times, where a 386 was still highend, this solution came as called. The only thing to do, was to load the samples on the card. Henceforth, one only provided the chip with playback commands, and this took care of the rest.
While playback quality was on CD level, the reception was was only possible at 8 bits, unfortunately. However Gravis offered an add on board, that also removed this deficiency. Titled as 16-bits upgrade module, it was not only this - in fact it was a whole Crystal-DSP with the ability to work with 48 Khz - the same, that was later standard on the Gravis Ultrasound Max (1994, 512K Wavetable).
This all sounded too beautiful to be true, if there was not one fault: the card neither possessed an OPL-Chip nor the ability to emulate a Soundblaster by 100 percent.
So in AdLib mode, the GF1 took over the role of the OPL2 together with a few Samples, and the software provided compatibility to the Soundblaster.
Despite that this was largely stressed, it didn't run usually, or only with a lot of effort. Consequently, Gravis had created a new hardware platform, for which there was practically not any support initially. Therefore they went to game producers, published extensive programming tools together with source codes, and not only few game patches, to crank the support. So the stone came gradually into rolling, and finally the first games came on the market, that supported the chip directly, also by 1992-1993.
One can see the card already as direct fight announcement to Creative Labs, hence they were forced to improve their Sound Blaster line.
Samples
| Monkey Island 1 | | [OGG] [MP3] | | GUS in MT-32 emulation mode; standard patch | | Epic Pinball | | [OGG] [MP3] | | The strentgh of the GUS: MOD playback with interpolation | | Descent | | [OGG] [MP3] | | Descent I intro - the card mustn't hide |
|
Technical
The GF1 itself was a quite simple processor. It always ran at a fixed frequency in the loop, and so it calculated data continuously. Through the solid frequency and hence the fixed computing power, it happened that the sampling rate of 44,1 KHz under 16 used channels could sink down to 16 KHz when using the full 32 channels.
One fed it with mono data of 8 or 16 bits quality from the sample RAM, said whether to loop this or not, how fast and loud it should play, and that was all about it.
Then the chip mixed the data, while the its peculiarity was, that it interpolated it!
Instead of simply only stretching digital data, the chip calculated inter values. This way it gained the rendering quality that so many people were emphasizing. The topic "Interpolation" is explained more extensively in the glossary.
|
AdLib Gold /1000 /2000
At almost the same time and price tag of the SB Pro, AdLib offered more value for the money. The ADC/DAC worked with 12 bits, instead of only 8 like with the SB Pro, while the card could even be fed with 16 bits sound data.
Additionally, playback runs at 44,1 Khz and stereo - while playing mono, the card could even mix both channels!
An OPL3 was used for the music rendering, and enabled stereo with 20 voices with 4 operators maximum each - at the time of the appearance Creative still offered the first revision the SB Pro, consequently AdLib also was in advance to the competitor there.
A golden sparkling paint finish underlines its name
A big peculiarity was the extremely fine gradation of the mixer, which offered 128 volume steps per channel. An add-on board, which offered the card virtual Surround sound together with echo, represented another highlight. I never have heard it, however it is said to improve sound quality significantly.
There were actually also different variations on the market: while there was a version dedicated to IBM's Microchannel ("2000 MC"), there also existed the Gold 2000 for a 16 bits ISA slot, and added SCSI-functionality.
Decidedly, they could cut out Creative with this card easily from the technical side. Crystal clear sound quality thanks to multiple filter steps, bus transfers could be monitored no more involuntarily, a higher resolution as well as sampling frequency and nevertheless the same price.
However all this had a catch: also this card was not compatible with the Sound Blaster! Surely downward compatible with the own AdLib - but nothing more. Annoyingly, however, games wanted to see a Sound Blaster for sample playback. On the AdLib Gold, they therefore didn't work, supporting Games should follow later.
Another problem was the late appeareance on the market. Promises already came early, however this - even if outstanding - result something too late. Consequently, this card is indeed very rare, and very difficult to get.
Mediavision Pro AudioSpectrum /16 (1991/1992)
What made this card a good challenger, was its compatibility with the SB Pro, as well as the OPL3 for the music rendering, and finally its 44,1 Khz, which it also offered when handling stereo data.
The compatibility lifted it up from the at the same time appeared AdLib Gold, and additionally it offered a SCSI interface for CD-ROMs - an outstanding feature for that time!
Pretty short after the first version, they pushed the 16 bits version of the card, which was pricely oriented by the SB Pro - namely $300.
These qualities and at the same time the good value for the money helped the card to get a quite wide spread, and so also little later its native support by games, where it's often presented in the abbreviation "PAS-16."
1992 - 16 bits, finally also from Creative Labs, SCC-1
Sound Blaster 16
SB 16 ASP - thx to Jörg Weske
Full CD quality with 16 bits and 44,1 Khz was finally also possible at Creative Labs, and finally, they had also put effort to clean up with the chaos of the interface of the predecessors. Unfortunately, they've done their job very thoroughly.
Generally, the consumer thinks if a product carries the same brand, backwards compatibility should be given, and on the product packet there's also a "Sound Blaster compatible". Therefore everything should be ok, shouldn't it? Yes, because the imprint didn't lie... the card is actually SB-compatible - nonetheless, however, not SB Pro compatible! Admittedly the card plays along without complaints if you use it as SB Pro compatible, and returns also everything with the correct frequency. However, there was a little point, that they forgot in the pre-delivery inspection - the card could not play stereo while being in SB Pro mode! This manifested itself in a very twisted (mono) playback of stereo data, since data is sent consecutively encapsulated to the card, alterning left and right.
Creative will officially probably never call this a bug. It is however already peculiar funny if the card processes the rest of the SB Pro commands, and only fails at this point....
They seemingly also cheated in some other aspects. Namely, the ADC could dissolve only 12 bits! Many users could prove this doubt-freely in their attempts, however this has never been officially confirmed. That is even more tragic, as exactly these 12 bits were seen as the disadvantage of the AdLib Gold compared to this card.
Marketing is simply everything.
Nevertheless, this card found many buyers. After all also old games could often work easily with the card, and also upcoming games adapted this card very fast, thanks to Creative's support. So, also this card placed a quasi standard again, that could not be broken by anybody else. The integrated ASP of the more expensive models however was practically never used. Probably also, because they - contrary to the usual practice - never disclosed any direct programming guides.
Creative Labs Wave Blaster I
It shouldn't take too long that competition offered this extension interface on their cards, too...
Samples
| Silpheed | | [OGG] [MP3] | | Silpheed in the version of Quest Studios on DB50XG | | Rhizome | | [OGG] [MP3] | | DB50XG in its typical area - electronic music | | Descent | | [OGG] [MP3] | | Descent I intro on the DB50XG - even better than SCC-1 | | Descent | | [OGG] [MP3] | | Descent I intro on the Wave Blaster I - lousy |
|
Links
| Mobygames | Games supporting the Sound Blaster 16 |
|
Roland SCC-1
Another pure MIDI card of the house Roland was developed, this time based on the external Sound Canvas module SC-55. Also this one was primarily targeted at professional musicians, and represented the first PC card with Roland's expansion of the General MIDI standard, General Synthesizer (GS).
The card has got 24 channels, that were fed with 16 bits PCM samples from the 4 MB sized sample ROM, and were mixed with 44,1 Khz resolution. Also here is valid, as with the LAPC-1, that 24 channels doesn't mean 24 different instruments playing simultaneously. However it's not as extreme as with the LAPC-1, 16 Instruments could be played, 16 times multitimbral here. The ROM content was compressed and included 317 sounds in the first, 354 sounds in the second version of the card.
There was an MPU401 compatible interface on board as well, consequently it was supported by almost each side - the card was plugged in and did its work. No drivers, no unnecessary hurdles.
Additionally, the card offered chorus capability, in opposite to the LAPC-1 that only supported reverb/delay. The card could also be brought into MT-32 compatible mode, but no samples could be uploaded! Games doing this - and their amount was quite not little - didn't sound like on the original LAPC-1.
However you shouldn't go on the card this way, because it presumably represents until today the highest of the feelings when dealing with MIDI playback! It's valid also here: I shouldn't talk about it, you have to hear it!
As always, you buy good with Roland, but definitely not cheap. The card was prized at about $500 - approximately the same, as the LAPC-1, also still temporary sold parallel.
Trough this it is, as the LAPC-1, a rarity, and only hard vs. expensive to get in hands.
The RAP-10 should be mentioned as well, which approximately hit the market one year later. This was the first consumer card by Roland, which also mastered sampling on CD niveau. The MIDI part was with 128 instruments (based on the SC-7 module) in opposite to the SCC a bit shortened, but otherwise identical. Though this card was intended only for Windows use, it was supported by not a few games, either directly as RAP-10, but also indirectly (MPU401/SCC-1).
Samples
| Monkey Island 1 | | [OGG] [MP3] | | SCC-1 in MT-32 emulation mode | | Silpheed | | [OGG] [MP3] | | To SCC-1 converted version by Quest Studios | | Legend of Kyrandia | | [OGG] [MP3] | | This card brings sound to almost everything... | | Legend of Kyrandia 2 | | [OGG] [MP3] | | Closing Theme | | Midi Test | | [OGG] [MP3] | | Descent I intro |
|
1993 - The AWE32
AWE32 PnP equipped with 8 MB RAM
However the sound was so humble, they wisely took the board off the market and replaced it with the Waveblaster II - which incorporated the EMU 8000, a chip developed by E-mu systems Inc., shortly bought by Creative Labs. These boards added a real MPU401-Interface the the SB 16.
Since these boards did their purposes better than expected, why not integrate them on a single card? Creative did exactly this, and the result was called "Advanced Wave Effects". The "32" also served more, to tell the consumer the advanced technics. Of course, the DSP's of the card still worked with 16 bits, merely the EMU 8000 could process 32 voices simultaneously.
This wavetable synthesizer was connected to an own sample RAM and ROM, which in the basic equipment were sized 512K (RAM) and 1 MB (ROM - UNCOMPRESSED, contrary to many different statements). It mixed with a better form of the interpolation, that should minimize distortions, and masters Reverb and Chorus.
Only the full version of the AWE could be memory-expanded in form of SIMM's, owners of the Value Edition went out empty. Maximum possible was a 28 MB expansion. The chip of course could access 32 MB, however the first 4 MB were already reserved for the ROM.
An also interesting capability of the card was the possibility to capture sound data of the E-mu chip directly with CD quality as digital signal in form of S/PDIF.
However, it became even more interesting if one knew that Creative saved once again. In order not to have to accommodate another third DAC, they decided to connect the OPL3 directly digitally to the E-mu.
On one hand you could capture so even the OPL3 data digitally, on the other hand however this way they "stole" the chip two voices! The EMU 8k can therefore only play 30 voices on the AWE32. Through driver optimization, however, Creative succeeded to compensate this mainly.
More annoying the following: they saved the MPU401-interface also again and favored the proprietary AWE interface instead. They handled compatibility merely in form of a driver or better: emulator, which didn't work in some situations.
Explanation in the FAQ concerning that kind of problems: "As more developers include support in native mode, this NMI problem will gradually disappear"...
In the whole, the AWE admittedly was only a "better" SB16, however its tonal qualities could expel competitors at least in the same price league. And so the card represented the tonal climax of ISA cards of the house Creative Labs.
Samples
| Monkey Island 1 | | [OGG] [MP3] | | AWE32 in MT-32 emulation mode -- awful!! | | Descent | | [OGG] [MP3] | | Descent I intro - a bit too much of hall |
|
Links
| The Truth | The whole truth about the AWE - some interesting details | | Mobygames | Games supporting AWE32 |
|
1995 - The end of an era...
(Advanced) Gravis Ultrasound Plug'n'Play
Fire red GUS PnP with 2 MB RAM
For the purpose of compatibility it was switchable to GF1-mode via software, however with an obstacle, that is not farther tragic on the second glance. There were two versions: one without and one with 512K sample RAM (then named "Pro"). You remember: the aged Ultrasound could not work at all without RAM!
Independent from its version however, you could increase the sample RAM using 2 SIMM's to get up to 8 MB - and so nothing more hindered the compatibility mode.
Some resourceful crafters later found out that the DSP supported up to 16 MB - and soldered two further SIMM holders on the back of the card! Instructions regarding this topic can be found in the Web.
As with the predecessor this card mastered mixing of 32 canals in hardware - now however always on CD level! An addition were effects like echo, reverb, chorus, flanging and fading - but these presupposed installed RAM.
The digital recording and playback of sound data happens with the accustomed 48 KHz.
What some users drove smoothly into craziness was the high noise level of first card revesions, which was result of a design mistake of the InterWave chipset.
However, AMD pushed a new version of the chip quite fast, and owners with the previous revisions could exchange their cards gratuitously with Advanced Gravis. Nevertheless, they lost fast the interest to produce this chip or even to develop it further. Presumably, it was not worthwhile. Annoyingly, however also all cards based on it - consequently also the GUS PnP - were influenced by this.
So Advanced Gravis made a step backwards - equipped the "old" GUS with an SB Pro compatible chip made by ESS, which right away replaced the Crystal CoDec, named the result "Extreme" and threw it on the market. Unfortunately with only minor success. After this relapse, Advanced Gravis gave up the development and production of sound cards - "no more profitable business".
Conclusion
So the most interesting years of development were described. Everything that followed was neither earth-shaking nor had it anything in common with the world of DOS.
I hope I met the nerve of some readers, and have enriched the knowledge of some others. I couldn't count up the time, to collect and write everything, anymore.
Certainly there wasn't place for each detail, however the content seemed already so to gradually explode. Whoever would like to know more in each case, use the already named links, and have also a look at the following ones.
I estimate at this point, I should get rid of my thanksgivings, because this gigantic article possibly hardly would exist without the help of others!
Jim Leonard contributed a not unimportant part - he recorded not few and helped me out with a picture and a description of the MFC. As well, Eike Belgardt enriched the article about not few samples and pictures meanwhile. Thanks again!
Thanks should also fly to Arno Fleck and Johannes Bollmann. Both helped me with reading through to remove some of the last discrepancies.
I would like to thank furthermore Juho Sippola, Jörg Weske as well as Olivier Boisseau for the supply of photos - just because I admittedly don't possess everything mentioned here. Thanks should also fly to Christoph Becker. If this article together with samples should explode my free volume of my provider, I have webspace where I can redirect to.
For further questions or inspirations, I'll be available in the Forum
Sample overview for direct comparison
|
Background
This article goes back more or less also to my childhood. As I was 13 years old, I sat in front of the PC with friends at that time, and we occupied ourselves above all with computer games. What particularly catched my interest at that time were the many options like "Roland", "AdLib" and similar. I was not still conscious at all for myself at that time that there were also other possibilities of the ghastly tweeting from the PC speaker. With 15 years age, I finally had an AWE32 Value Edition.
Don't get misled from the title, this card was still far away from beeing cheap. It was prized at 340 Marks 1995 - however I finally could play everything with real sound!
Simultaneously, I held also a book named "PC Underground" in my hands. Not exactly outstanding, but filled with indications of the legendary Gravis Ultrasound and with scene demos.
During this time, I also learned to program the Sound Blaster. The book however quickly became an obstacle, because some information was completely incorrect, and so I got my information from somewhere else (my works for it still are available under the category "coding" on this page).
However I still didn't know what "Roland" was. But also the rest of the sound options were interesting for me, and I finally searched the Web, and I found indications of a "LAPC-1" by Roland. Despite my efforts, it was not possible for me to get such a card in normal ways.
And while I had possessed some other legends, like the Gravis Ultrasound, long ago meanwhile, I finally got my LAPC-1 in summer of 2002. Over this auction house, I had already gotten some other cards, and it still should do...
It is therefore not one of the cheapest hobbies, as you probably can imagine. Nevertheless, the possibilities, that you already had in these times, fascinated me - only they were quite exorbitant for me as teenager. And as these cards shouldn't collect dust in my wardrobe and remain unseen for years, they should now be described and shown as good as possible here.
Glossary
Sampling frequency
The frequency of the retrieval of a signal while conversion of the analogous signal into a digital one. A CD possesses the often named frequency of 44,1 KHz or 44100 Hz, for example.
ADC - "Analog to Digital Converter"
A component, which performs the conversion of an analogous signal to digital data. Its quality is mainly determined by sampling frequency and resolution.
Attack/Decay/Sustain/Release
These concepts mirror a split up of the behavior of a real instrument in order to be able to reproduce such a one in the world of synthesizers.
"Attack" stands for the attack of the Instrument - the part, where the sound reaches its maximum volume,
"Decay" for the time period of the sound reduction, until it meets with the pre-defined volume of the sustain,
"Sustain" is the time period, in which the sound remains at the same volume level,
"Release" determines the time period, in which the instrument fades out
Resolution
While the sampling frequency represents the frequency of the signal retrieval, the resolution defines the fineness, in which the amplitude is measured.
Usual resolutions are 16 bits with the CD, while Dolby Digital (AC3) usually works with 18 bits and dts usually works with 20 bits. Highend sound cards even offer 24 bits meanwhile.
DAC - "Digital to Analog Converter"
Almost the same as the ADC, with the simple difference of reversed direction
Demo scene
In the beginnings of the computer time, a type of scene developed - people interested in computers, who understood something about hardware programming. In the beginning almost only active with Commodore machines, interest went over to the PC soon.
They had sat down with the goal to squeeze out the maximum from the existing hardware, and above all in Finland and round there are meetings and contests still today, in which people show their proficiency.
However, at the latest in '95 the actual hype eased, since almost everything was explored, that was to explore. Moreover the technology was so efficient meanwhile that it didn't represent any more big challenge.
More information at: ,
DSP - "Digital Signal Processor"
A processor specialized in a certain waveform, in our case for the processing of sound data
FM synthesis
The sound of a natural instrument consists, splitted into its components, of a main wave - the sound, which is the most evident - and several harmonic waves. The latter define the character of an instrument.
This theory was set up by John Chowning, student of the Stanford university, in the year 1972, and so many synthesizers of the company Yamaha do it in practice.
More on the history:
GF1
The of the Gravis Ultrasound and license products used DSP. It can mix up to 32 channels at a maximum of 44,1 Khz and 16 bits.
Interpolation
If digital information, because a sample should be played with half speed for example, is stretched, so a type of stairway effect originates. These stairways can be seen as a a small rectangle signal - at their edges emerge higher interfering frequencies, which are dependent from the actual frequency.
In order to counteract this effect, we calculate inter steps to fill the gaps produced by stretching, to minimize interferences. Hence an interpolated signal sounds essentially more natural.
The above displayed example only shows an ideal interpolation. However in reality, it rather looks like a point-to-point connection.
MIDI - "Musical Instrument Digital Interface"
Protocol, which was intended to record and return digital music on digital synthesizers. Initially, it was merely intended to interconnect two keyboards. However the technology was quickly adapted to computers.
The technology doesn't transport digital sample information, as for example PCM samples, but merely which instrument was used, in which pitch, how long it plays and the attack strenth etc.
Soon a standard called "General MIDI" established, which defined a standard set of instruments, primarily based upon Rolands own definitions.
MPU-401
Also coming from Roland, this is a MIDI interface grown to standard in the world of PC. If a software offers MIDI support, it will primarily support this standard.
Multitimbral
In contrast to the possible channels, which a card can mix in hardware simultaneously, this explains the number of instruments, that a card can return simultaneously. This is to be explained the way that an instrument can take more than one channel to play.
OPL2/3
The Yamaha OPL series are in versions 2 and 3 pure FM synthesizers, while version 4 already offered Wavetable synthesing. All models are downward compatible.
OPL2 offers mono sound with 6 to 9 voices and 2 operators, OPL3 already stereo with 20 voices and 4 operators maximum.
S/PDIF - "Sony/Philips Digital InterFace"
A standard that specifies the data exchange of digital sound data. Used in the beginning primarily for DAT recorders in order to avoid the detour over analog signals and hence to allow a copy reproduction bit by bit.
Wavetable synthesis
For MIDI playback used technology, so to speak the counterpart to FM synthesis. Real recorded instruments are stored in memory and are played slower of faster according to the wished pitch.
This technology could meanwhile oust FM synthesis considerable in the computer area.
|
