WaveFrame Corporation – AudioFrame

This article is dedicated to this extremely rare machine.

The aim here is to share information about this exceptional machine, but also to serve as a showcase in order, I hope, to make contact with former WaveFrame employees, as well as former users or owners.
Indeed, I have already taken this approach for other brands (Fairlight, Ned, Technos, etc.), and it has always encouraged the collection and sharing of information and experiences. My ambition? To bring this machine back to life.
If you are a former WaveFrame employee, a user from the time, or if you still own one of these machines and have access to documents (user manuals, technical manuals, software, hardware), please do not hesitate to share them: this would help keep this mythical brand alive, or bring it back to life.

But first, I would like to thank UVI here, and especially its director and co-founder, Alain Etchard, who gave me this rarity. Having worked with them in the past on several of their products (darklight-iix, the-beast, energy, synthox and u1250), they thought of me…

The sounds of this machine are also available from UVI in their software package: px-waveframe

Let us go back in time: this machine is one of the great mythical machines of the late 1980s, a kind of grail of what was best in audio synthesis at that time. I had seen its name in articles, I no longer remember whether I had seen photos, but one thing is certain: it remained even more mysterious than the Fairlight and Synclavier systems I had already encountered on several occasions, and for several years, in studios around the Paris region. Which is quite surprising given that it was marketed much later, from 1988 onward,

It is also an extremely expensive machine, whose purchase price was around US$88,000. I also found a trace of a price of DM100,000 around 1989.

Probably one of the first versions of the DAR, the front panel is different.

And one day, I came across it, quite some time ago now, it must have been at least 26 years ago, at Funky Junk by chance, while going there to test and validate the proper operation of a CMI III, the one belonging to Celmar Engel (repainted purple with glitter… apparently, his wife wanted it that way) which was on consignment sale. It was at a time when Funky Junk still had a consignment showroom in France, around La Dรฉfense, near Paris.
It was in very good condition, and above all in a complete configuration, that is to say with the large black unit (the same one that UVI gave me), but also a complete PC computer, the documentation (binders and floppy disks) and cables.

Typical configuration: AudioFrame “DAR” + PC video monitor running MS Windows + keyboard and mouse.

AudioFrame

Developed in the late 1980s by WaveFrame, the AudioFrame stands out as a highly advanced digital audio workstation for its time. Its design is based on a modular architecture organized around a rack-mount chassis and a high-speed backplane capable of handling up to 64 multiplexed audio channels. This approach makes it possible to build a fully configurable system by adding dedicated boards for the various functions: audio conversion, DSP processing, sampling, multitrack recording, or synchronization interfaces. One of the major strengths of this architecture is its scalability: it is possible to adapt the machine precisely to a studioโ€™s needs, then expand it progressively without calling the entire system into question.

From an audio standpoint, the AudioFrame efficiently combines sampling and real-time processing functions. The sampling modules each provide up to 16 dynamic voices, with flexible resource management and possible expansion through the addition of memory or extra modules. The overall polyphony therefore depends directly on the installed configuration, which makes it possible to achieve significant capabilities for the period. Processing is handled by DSP boards incorporating several specialized processors, capable of carrying out real-time mixing, equalization, and effects, with 24-bit internal processing ensuring high audio quality.

The AudioFrame also stands out for its hard-disk multitrack recording capabilities. Thanks to its dedicated modules, it can record and play back several audio tracks simultaneously with non-destructive operation, precise edit points, and flexible take management. This approach makes it a true advanced digital multitrack recorder, particularly suited to production and post-production environments. Total recording time can reach several hours depending on the configuration, which was remarkable at the end of the 1980s.

The storage chain is based mainly on SCSI hard drives integrated into the rack or connected through expansion units. These drives make it possible to store large quantities of audio data and ensure fast transfers to the memory of the sampling modules or to the multitrack playback engines. Loading times are thus greatly reduced, which considerably improves workflow. The addition of expansion racks makes it possible to increase storage capacity even further, confirming the systemโ€™s orientation toward intensive use in a professional studio environment.

Another essential aspect of the AudioFrame is its synchronization capability and its integration into complex environments. The system supports professional standards such as SMPTE (linear timecode), VITC for video, as well as external synchronization signals (word clock, house sync, etc.). Dedicated inputs and outputs make it possible to lock the AudioFrame precisely to external sources or to drive other equipment. At the same time, several MIDI interfaces are available, with independent inputs and outputs, making it possible to control external instruments or integrate the system into a broader music production chain. All these signals circulate at high speed inside the system, which largely eliminates the classic limitations of MIDI used externally.

Finally, the AudioFrame stands out for its overall level of integration. It is not simply a collection of independent modules, but a coherent platform in which sampling, recording, processing, synchronization, and data management are closely interconnected. This convergence, combined with a modular architecture and substantial processing power, makes the AudioFrame a direct precursor of modern digital audio workstations, while retaining a hardware-oriented approach typical of high-end systems from the late 1980s.

The AudioFrame consists of a large black 19-inch 10U rack unit (height) and is very deep. Here are the external dimensions of the enclosure: 48.5 x 44.3 x 64 cm, and it is heavy, I would say more than 40 kg. This large enclosure is called the DAR (Digital Audio Rack).

The front panel and its power button, the 3 activity LEDs, one of the hard drives, as well as the tape drive
The rear, with the 10 slots that actually house 12 boards.
The rear of an AudioFrame.
The rear of my AudioFrame (before cleaning).
The rear of my AudioFrame (after cleaning, yes I swapped the two sampling blocks).

Inside this enclosure are grouped all the electronic boards (also large-format, multilayer boards: 36 x 28 cm) and three hard drives (in the configuration I have: 2 drives of 600 MB and one of 2.1 GB), as well as a magnetic tape drive/recorder (for backing up the contents of the hard drives). All of this works in SCSI.

The rear interior of my AudioFrame.

It also contains what is needed to supply power to the whole system:

  • a switching power supply for the 5 V rail,
  • a linear power supply (large transformer, rectifier bridges and voltage regulators) for the voltages dedicated to the analog section.

But in order to operate, it also has to be associated with a PC-type computer. You have probably understood that I did not have that famous PC, nor do I have the software and the documentation.

Apparently, it was the ALR brand (Advanced Logic Research) that was configured and delivered by WaveFrame. However, in my case, since I do not have the original machine, I think that a PC equipped with an x86-compatible processor (ideally a model between the i80386 and the Pentium), configured with the Windows operating system and fitted with a Token Ring-compatible network card, should have been sufficient.

According to the information I have gathered, the control software would run under Microsoft Windows 2 up to 3.11.

As the logic of this very old family of Windows operating systems required (from version 2 and those that followed until the arrival of NT and 95), there were strong constraints that required splitting the system into several programs. It is therefore very likely that there were several programs, each responsible for a specific area of management (MIDI, hard-disk recording, sampling, effects, etc.).

Here are a few screenshots of these program interfaces:


SCP Module

Studio Control Processor

This basic module handles all MIDI and timecode control. It reads and generates LTC (SMPTE), reads VITC and composite sync, links the computer to the Digital Audio Rock (DAR) and, if necessary, to several DAR units at the digital level. It also allows the AudioFrame remote control to be connected.


SAMPLING Module

It is possible to integrate up to 3 modules into a DAR, each with 16 dynamic voices and a constant sampling rate of 44.1 kHz. No anti-aliasing filter is therefore required. Each sample immediately delivers a perfectly pure sound across 14 octaves, without any digital noise. Each sampling module can manage up to 28 MB of RAM and handle up to 614,400 samples.

My machine is equipped with two modules and totals 60 MB of RAM.

DSP Module


DRM-4/8 Module

(Hard-disk recording module)

The Digital Audio Rock can accommodate up to 4 DRM modules.

Wave From therefore offers the worldโ€™s only hard-disk recording system currently capable of 24-bit recordings.

The usual limitations of hard-disk recording systems are completely eliminated with the AudioFrame:

  • unlimited crossfade durations
  • unlimited number of tracks (thanks to the digital networking of several DAR units)
  • unlimited number of cue points (direct selection without rewinding), variable speed at 100%, etc.
  • Current maximum recording time per track: 7 hours (!)
  • Free distribution of recording time among mono, stereo, or multichannel events.

The following fact illustrates the capabilities of the AudioFrame particularly well:

the DRM module plays 50% more tracks when there are no long crossfades. A DRM-8 is therefore a complete 8-track editing and recording system, but without crossfades it becomes a 12-track machine.


ALR-Computer System

DIGITAL-TO-ANALOG MODULE – 5131 DAC

DA converter, outputs

It is possible to plug in as many sipd modules as desired. Automatic calibration, as is already the case for the AD converters, guarantees that all channels sound the same! (16 bits, 44.1 kHz, 20 Hz to 20 kHz, 96 dB.) In addition, this board includes 8 digital outputs that are available in parallel and simultaneously with the analog outputs (constant sampling rate of 44.1 kHz).

The boards

Since the DAR operates with a BUS, it is possible to configure the machine for a use more oriented toward recording or sampling… by adjusting the number of modules in the enclosure. It is stated that it is possible, for example, to install 3 SAMPLER modules (which apparently also implies the same number of MEMORY modules, thus representing 6 modules in total), but it is also possible to install 4 DRM modules.

I found variants among the modules: indeed, 2 versions of the ADC module, and 3 versions of the MEMORY module.

5101 ADC – 2 audio inputs in XLR format

5101 ADC

5102 ADC – 8 audio inputs in XLR format, switch (ENABLE/DISABLE)

5102 ADC

5203 UDI-4

5203 UDI-4

5131 DAC – 8 audio outputs in XLR format, digital out in female DB9 format, DITHER switch (ON/OFF))

5131 DAC

5501 SCP

5501 SCP
5501 SCP

5601 DRM – RS-422 in male DB9 format, SCSI A and SCSI B in male DB25 format, DSP A in female DB9 format, DSP B in male DB9 format

5601 DRM
5801 DRM

5711 MEMORY – SCSI 1 and 2 opening in male DB25 format

5711 MEMORY

5712 MEMORY – SCSI 1 and 2 opening but without connector

The module is made up of two stacked boards attached to a common front panel, and in addition it is connected to the SAMPLER module by two ribbon cables.

5713 MEMORY
The 224 KM41C1000 RAM chips (1Mx1bit), representing a total of 28 MB
CPU i80186, RAM: 32xMT1259 = 1 MB, ROM: 2x27C64 = 16 KB, WD33C93, D71071, Z8530APC

5713 MEMORY

5713 MEMORY

5801 DSP – IN in male DB25 format, OUT in female DB25 format

5801 DSP

5901 SAMPLER

The module is connected to the MEMORY module by two ribbon cables.

5901 SAMPLER
CPU i80186, RAM: 64xMT1259 = 2 MB, 2 memory sticks: 512 KB, ROM: 2×2732 = 16 KB, Z8530AB1

Sources on the web

For years, I searched the web for information, documents, and media about this mythical machine… I have always been wary of the comments and opinions of people who labelled themselves specialists, and that also applies to what I have read in magazines and on the web. I am not even mentioning the level of trust one can place in forums and various websites published online, which far too often come close to the level of competence of conversations in a roadside cafรฉ at rush hour, early in the morning…

Moreover, I kept coming across the same information over and over again, more or less distorted, along with a few rare websites, and nothing more… It is always tedious, even very frustrating, to find nothing when digging into the depths of the web, where one searches every tiny lead that often goes nowhere: that is where one sees the true limits of digitization, because only what has been digitized is available!

I have always been relentless in my efforts to obtain information. I have this need to satisfy my knowledge on a subject or in a field. I have had this character trait since my distant childhood, and it is rather difficult for those around me, because it quite often turns into obsession. And often, if I think I have recovered as much as it was possible to extract from what was available at that time, I put it away in a corner of my brain, then move on to something else, another objective, another target…

A natural source of information would have been the WaveFrame brand website, but unfortunately,

Fortunately, thanks to Web Archive, it is possible to go back in time, but here that is not enough… Too bad:

https://web.archive.org/web/20260000000000*/http://www.waveframe.com

I also found a few traces of user opinions and feedback:

In English:

https://gearspace.com/board/showpost.php?p=12470801&postcount=51 (February 27, 2017)

My submission is the 1987-era Boulder Colorado-based WaveFrame AudioFrame.
I received maintenance training on the AudioFrame 1000 sometime around 1990. I had several clients in Dallas using them including Charlie Pride’s studio and Real to Reel which did a lot of radio production on them. Many of the Motel 6 “We’ll leave the light on for you” spots with Tom Bodett were done on the WaveFrame.
The LAN adapter used a black box interface to an IBM Token Ring network. Drive interfaces were SCSI and the original ones if I remember correctly ran the user interface on Windows 2. The DSP was Motorola 56K.

A restoration project for a more modern model: the WaveFrame 401.

Bringing an iconic DAW back to life – The Waveframe 401

In German:

https://www.sequencer.de/synthesizer/threads/waveframe-1000-audioframe-1000-system-noch-brauchbar.115602/ (July 6, 2016)

Audio examples

And through all these publications, I compiled a list of the named users and their productions: Michael Cretu (ENIGMA, album MCMXC a. D. – 1990), Peter Gabriel (Passion – 1989)

I also found two promotional booklets, one in English and the other in German. They are not perfect, some pages seem to be missing, and the full surface of the documents was not digitized.

An article in Keyboard Magazine US:

An article in Sound and Sounds:


I made an inventory of the machine:

12 large-format boards ()

6 Intel 80186 CPUs [Intel and Siemens] (I know this x86 family well, I even wrote a book on their assembler https://www.amazon.fr/Assembleur-x86…/dp/2744015644…), 6 Motorola XSP56001 DSPs, 8 MB of RAM for the CPUs, 128 KB of ROM for the CPUs, 56 MB of RAM for the sampling boards, 4x 16-bit 50 kHz stereo ADCs (easily expandable to 8x), 4x 16-bit 50 kHz DACs, tons of digital and serial I/O, and 4 SCSI controllers… All distributed across 12 giant 28×36 cm boards.

The RAM for audio represents 28 megawords of 16 bits. With sampling at 50 kHz, this represents a sampling time of 560 seconds (about 9.33 minutes) in stereo.

If we apply this reasoning to the three SCSI disks, one of 2.1 GB and two of 600 MB each: In simple stereo: 2 hours and 24 minutes of recording possible. In multitrack mode (8 stereo tracks): 17 minutes of recording possible. And in RAM: 2.45 minutes.

I forgot to mention that it is the former machine of Ian Robertson Underwood: https://www.imdb.com/fr/name/nm0881005/#AudioFrame#WaveFrame

The WaveFrame AudioFrame seems to have found its place mainly in the world of film and post-production, much more than in mainstream music. Among the few notable exceptions in music, we find:

– Peter Gabriel, who used it on Passion (the soundtrack to The Last Temptation of Christ, 1989) and on Us (1992), one of his studio albums.

– Michael Cretu (Enigma), who made it a key element of his iconic album MCMXC a.D. (1990), notably for its ethereal sound textures and transformed choirs.

– Stevie Wonder, who used it for post-production and sound creation projects between the late 1980s and the mid-1990s.

Why such a predominance in film?

The AudioFrame was particularly appreciated in post-production studios (sound effects, sound design for film and television). Its ability to handle long, high-quality 16-bit samples, a rarity at the time, made it an ideal tool for creating complex sound atmospheres and detailed audio landscapes, long before these features became available on consumer samplers.

Rodolphe Honestly… This is only the second one I have ever come across (the first was at least 20 years ago at Funky Junk in La Dรฉfense), and clearly, it really is a rarity (worse than Fairlight and Ned or Technos): almost no documentation on the net (a few things remain from the former manufacturerโ€™s site via Web Archive).

But for now, I still have too many things to finish: I am running a tremendous, even staggering, backlog in my workshop. So it will have to wait. For the moment, I have no opinion to give, except that it weighs as much as a dead donkey, or even much more 🙂

This instrument was designed to compete with the Synclavier, the Fairlight Series III, and the MFX. In 1988, a 48-voice system cost US$88,000. The sound quality is exceptional and sample editing is made easy by a very intuitive graphical interface running on a Windows PC to control the system. The sound quality far surpasses that of my Akai and E-mu samplers. It has a truly fantastic-sounding reverb. With the DSP board and the HD recorder options, the AudioFrame is a self-contained studio in a box, offering sampling, effects, MIDI sequencing, mixing with EQ, mix automation, and 24-bit hard-disk recording functions. The AudioFrame was used for sound effects and ADR in the film industry well into the 1990s. Peter Gabriel mentions the AudioFrame in the credits of โ€œUSโ€ and its most famous musical use is probably on Enigmaโ€™s MCMXC album.

Waveframe is still active and continues to manufacture digital audio workstations (DAWs), but like Fairlight, they no longer make samplers.

Comments on the sounds:

It is a sampler: if you feed it poor-quality data, you get poor-quality results. It comes with a fairly large sound library created by Prosonus, including many excellent acoustic sounds.

Sound on the NeXT

I have always had a keen interest in sound and the devices that generate it.

I discovered NeXT computers in the magazine Science et Vie Micro (SVM) in the double issue 63 of July-August 1989 1.

But the first time I saw and touched NeXTcomputers and NeXTstations, I believe it was at “Imago” (I’m not sure of the name) on Boulevard Saint-Germain in the 6th arrondissement of Paris (or at the SICOB show in 1990). At that time, this store was one of the few, if not the only, store in Paris that exclusively sold NeXT equipment. These machines were a shock !

They were beautiful, with an elegant design and that black color that stood out from the standards of the time, which were dominated by shades of beige and gray. And above all, they were powerful, combining the robustness and power of Unix with a superb and fast graphical interface. Not to mention the innovative interest of their development tools and the object-oriented approach of the system.


One of my NeXT Cubes

My desires came to a screeching halt… at the time, I absolutely didn’t have the budget to equip myself at these prices2…. For example, the two basic models were :

NeXTcube (68040) with 8 MB RAM, 105 MB disk, MegaPixel screen, Starting Point kit (keyboard, mouse, etc.), 2.88 MB floppy drive and NeXTSTEP 2: 50,845 francs ex VAT.

NeXTstation with 8 MB RAM, 105 MB hard disk, MegaPixel screen, Starting Point kit (keyboard, mouse, etc.), 2.88 MB floppy drive and NeXTSTEP 2: 31,765 francs ex VAT.

What’s more, if you needed to develop, you couldn’t neglect (to a lesser extent) the price of development tools and documentation, and you had to allow for more memory, a change of hard disk and RAM, with a maximum of 2GB (per partition) and 64MB RAM respectively.

So, reasonably, I waited… a long time before acquiring one (thanks again to the donation from my former colleague and friend Thierry Besanรงon)! What followed was a cascade of lucky events, including, among other things, the acquisition (by exchanging equipment) of two NeXTstations (B/W), two NeXTcomputers (cubes), and a multitude of peripherals, each more varied than the next! But very clearly, my machines come from the world of research and academia, in music, mathematics, and medicine.

Since I acquired these NeXT computers, and more particularly the NeXTcubes (commonly called Cubes), I have been almost archaeologically interested in everything that allows sound processing via these machines… and the associated research tools… And yes, you have probably noticed that we are gently drifting towards IRCAM.

A few cubes at home

But first, I propose a little trip back in time. Let’s take a jump into the past… And it won’t be about sound. Indeed, everything started at the end of Steve Jobs’ first period at Apple Computer… In 1985, Steve Jobs was ousted from Apple.

In 1986, Steve Jobs bought The Graphic Group from Georges Lucas for $10 million… and renamed it Pixar.

Does Steve Jobs like cubes?

In its early days, Pixar was dedicated to selling very powerful machines (200 times faster than a DEC VAX-11/780 computer) for the time, specializing in the creation and graphical visualization of data. These machines, known as Pixar Image Computers (PIC), were intended for various companies and laboratories with advanced graphical needs. Potential clients included hospitals (for three-dimensional visualization of X-rays), the aerospace industry, and even Disney, which used these computers to automate the coloring of its animated films.

However, both versions of the Pixar Image Computers were expensive ($135,000) and struggled to find buyers. To demonstrate the exceptional capabilities of their machines, Pixar decided to produce small demo animations. Among these creations was Luxo Jr, a short film directed by John Lasseter. This promotional film featured a small desk lamp, which has since become the emblem of Pixar. Luxo Jr not only succeeded in drawing attention to Pixar’s technological capabilities but also won an Oscar, thereby enhancing the company’s reputation in the field of animation.

Despite an Oscar, critical success, and technical recognition, Pixar’s machine sales did not take off and remained disappointing. In response, Pixar reinvented itself and started producing numerous advertisements, using their animation expertise to create memorable campaigns.

Pixar Image Computer I
Pixar Image Computer II
Pixar Image Computer II

I recommend visiting this site because they explain a lot about this mysterious machine. There is also a PDF version of the Pixar Image Computer brochure3.

Pixar sold 300 PIC computers from 1986 until it closed its hardware business in 1990. The division dedicated to designing and manufacturing machines was abandoned and sold to Vicom Systems, marking a turning point for the company.

Many of these computers were sold with custom software, enabling, for example, the processing of images from scanners in hospitals or Disney Animation’s digital ink and paint system. Some of them were used by clients until the late 1990s.

Paul G. Allen the co-founder of Microsoft has put his vast collection up for auction at Christie’s, which is like a world tour of computers, ranging from small CPUs to mainframes and even supercomputers. Among other things, he is selling a P.I.C.

NeXT

In parallel with his activities at Pixar, and likely to overshadow Apple, Steve Jobs launched a new company: NeXT.

The goal of this company was to design a new type of computer that would straddle the segment of workstations and high-end personal computers! Jobs wanted the power of Unix and an ergonomic and user-friendly graphical interface.

Indeed, NeXTstep4 5, the operating system of NeXT, is a WYSIWYG (What You See Is What You Get) type, meaning literally in French “ce que vous voyez est ce que vous obtenez” or more concisely “tel affichage, tel rรฉsultat” !

Since Steve Jobs was a perfectionist, he needed a computer offering high-end capabilities and a revolutionary design ! NeXT was born !

Steve Jobs et un NeXTComputer
Steve Jobs, the ebullient founder of NeXT and his first creation

At NeXT, there are only two types of machines: the cubes (NeXTcube6) and the stations (NeXTstations7, NeXTstations Color, and NeXTstations Turbo Color8). Of course, there are nuances between these different models…

We often talk about the Cubes, but in reality, we should focus on the CPU boards of the Cubes, as there were several versions. Initially, these machines were equipped with Motorola 68030 processors running at 25 MHz. Later, updates were offered to improve performance, with Motorola 68040 processors at 25 MHz, then at 33 MHz, and finally an upgrade to 50 MHz. These improvements also involved changing the motherboard to increase the RAM capacity, which went from a maximum of 64 MB at the beginning to 128 MB at the end. There were also variations in the BUS/Power backplane PCB.

To deepen your historical knowledge and get a more serious description of these machines, see the excellent site by ร‰ric Lรฉvรฉnez, which is undoubtedly the best French site (indeed, let’s be modest, on the web) about NeXT and in French! And also the must-see site/forum nextcomputers.org.

A NeXTcomputer that over time became NeXTcube (aka the “cube”).
NeXTstation
A black-and-white NeXTstation (known as “slabs”)
NeXTstation
My main black-and-white NeXTstation (known as “slabs”)
NeXTstationcolor
A color NeXTstation (called “slabs”)

It’s worth noting that the computers produced by NeXT (now known as “black hardware”) come standard with what was available as a very expensive option from competing manufacturers at the time. For example, all NeXT motherboards come with a Motorola 56001 Digital Signal Processor (DSP). This processor specializes in digital signal processing, and is ideally suited to sound processing !

dsp-56001
A Digital Signal Processor (DSP), specifically the Motorola XSP56001 running at 20MHz

The internal sound capabilities of NeXT computers

As mentioned in the previous paragraph, the NeXTs are equipped with a DSP operating at 25MHz and with 24KB of memory, expandable to 96KB (we’re not kidding!). This additional processor was the icing on the cake… These machines had some impressive features for 1991!

With two serial ports compatible with standard MIDI interfaces, a 16-bit mono or stereo sound acquisition and restitution system with a maximum frequency of 44100 Khz, these machines were ahead of their time! There were also in the development tools the “Music Kitโ„ข” and “Sound Kitโ„ข” specialized libraries to manage audio, DSP and MIDI events! It’s worth noting that the DSP has a DB15-format input/output port on the motherboard, allowing external cards to be added.

The functional structure of NeXT’s software libraries MusicKit and SndKit


The characteristics of these machines (especially the cubes) naturally suited them for use in a musical environment. Unfortunately, due to their prohibitive price and the limited popularization of Music and Audio Technology (M.A.T.) at the time, these remarkable machines were only used within research centers, notably the two largest: IRCAM in Paris, France, and CCRMA at Stanford in the USA. They were also utilized in other research centers in the Netherlands, Germany, Belgium, South Korea, Japan, etc.

NeXT computers have a DB19 port which serves as the black-and-white video output. As you will read in a few paragraphs about the Sound Box, this port is also used for connecting these devices. It’s important to understand that this port handles signals for keyboards/mice, audio, and also black-and-white video.

In addition to the native audio capabilities via the DB19 port and the serial and DSP ports, there are three ways to expand the capabilities of NeXT computers: internally, by adding a NeXTbus card, or externally, via the DB15 connectors of the DSP, and surprisingly, through the SCSI port.

Natively, NeXT cubes and stations are capable of managing sound at the system software level; different keyboard versions are equipped with specific keys (increase, decrease, mute).

First version of the non-ADB keyboard (the circle with a line through it on the Power key is different from the second version). The keys dedicated to sound (speaker with waves) control the volume. You can also see the green “Mute” label, activated by holding down the [command] key.

Second version of non-ADB keyboard (the barred round symbol on the Power key is different from the 1st version).

The only version of ADB keyboard, with round buttons replacing keys and the green power button.

Sound on monitors ?

The black and white MegaPixel Display N4000, N4000A, and N4000B monitors, in addition to their display and input functions (keyboard and mouse), are equipped with a speaker and microphone. At the back of the monitor, there are RCA connectors for stereo audio output, as well as two 3.5mm jacks: one for audio output to headphones and the other for connecting an external microphone. However, due to space constraints inside the monitors and to avoid display interference, the color monitors, which are not specific to NeXT machines, are not equipped with integrated speakers and microphone.

Connectors of an N4000 monitor
Connecteurs d'un moniteur N4000A
Connectors of an N4000A monitor
Connectors of an N4000B (ADB) monitor

Sound Box

To add sound to NeXT machines (Cube with a single color monitor connected to a NeXT Dimension, NeXT Station Color, and Turbo Color) equipped with color monitors, you need to use the appropriate Sound Box. There are two versions: the N4004 for early NeXT machines equipped with non-ADB keyboards and mice, and the N4004A for Turbo machines equipped with ADB. The box is connected to the NeXT via the DB19 port. This port handles signals for keyboards/mice, audio, and also black-and-white video.

Sound box no ADB

Sound Box N4004
Sound Box N4004A (ADB)

NeXT-External CD-ROM Drive

This is the CD-ROM drive offered by NeXT. As you can see, it’s a fairly typical SCSI CD drive for its time, and it uses a cartridge (caddy). You can plug in headphones into the front 3.5mm jack, and there’s a volume control knob right next to it. For listening through speakers, there are two RCA jacks (red and white) at the back of the unit. Interestingly, from the outside, the NeXT-External CD-ROM Drive9 closely resembles (the difference being color and logo) the AppleCD 150 marketed by Apple during the same period, which integrated a SONY CDU-514-25 CD Caddy drive.

Lecteur de CD

External audio devices

These external peripherals are one of the only possible upgrades, and are particularly useful for NeXT stations which, due to their physical format, are not expandable (as they have no NeXTbus). They plug directly into the NeXT DSP port.

Ariel – DM-N Digital Microphone (595$)


Ariel – ProPort Model 656 (1295$)

This external box is equipped with two analog inputs and outputs, pre-amplifiers and two phantom-powered inputs. Sampling frequency is adjustable from 8 kHz up to 96 kHz in 16-bit.


The Singular Solutions – A/D64x Audio Interface (1295$)

The Singular Solutions A/D64x system includes the hardware and software required for professional direct-to-disk recording and data capture on NeXT computers. The system features two channels of 16-bit delta-sigma (ultralinear) analog-to-digital conversion with 64x oversampling and a three-stage linear-phase digital anti-aliasing filter. It features balanced and unbalanced inputs and an integrated low-noise mic preamp with 48-volt phantom power. It also offers digital audio input and output (AES/EBU and S/PDIF).

The A/D64x supports sampling at 16KHz 22.05KHz 32KHz 44.1KHz and 48KHz (plus external). It can also operate in stand-alone mode (analog input to digital output). Multi-A/D64x synchronization is available, as are brackets for rack mounting. Software is included for direct-to-disk recording and non-destructive audio editing.


Ariel – DatPort AES/EBU CP340 Digital Audio < -> DSP Port Interface (??? $)

This is a digital-only audio interface IN / OUT, XLR, RCA, Optical that operates in mono or stereo at frequencies of 32, 44.1 and 48 kHz. Communicates with NeXT via DSP port.


Ariel DAT-Link+ Townshend Computer Tools, Inc

It’s a digital-only audio interface IN / OUT in AES/EBU, SPDIF RCA, Optical and works in mono or stereo at 32, 44.1, and 48 KHz frequencies. It’s truly astonishing, since communication with the computer is via the SCSI port for workstations of the time (sun, sgi, dec, pc), but also on NeXT.


MetaResearch – Digital Ears (595$)

MetaResearch โ€“ Digital Ears
MetaResearch โ€“ Digital Ears

The Digital Ears takes line-level audio signals (rca connectors) and converts them to digital information in 8 or 16 bit, 5.512KHz, 11.025KHz, 22.05, 44.1 kHz stereo or 88.2KHz in mono. It then transmits this information to the digital signal processor (DSP) on the NeXT Computer.


Internal computation board

I’d like to make it clear right away that I’m talking about “computing boards” here. In fact, in their basic version, or in the absence of converters (analog to digital and digital to analog), these cards do not allow you to produce sound directly!

To the best of my knowledge, there are only two models available, namely IRCAM‘s M860 and CCRMA’s Quint Processor, both built in cooperation with the Ariel company.

Ariel / IRCAM – M860 (15000$)

m860a
IRCAM / Ariel – A full-option M860
M860
IRCAM / Ariel – An M860 without options
IRCAM / Ariel - M860 board without heat sinks, RAM and
IRCAM / Ariel – M860 board without heat sinks, RAM and piggyboard
Zoom in IRCAM / Ariel - M860 board without RAM and piggyboard
Zoom in IRCAM / Ariel – M860 board without RAM and piggyboard
piggy board
To the left vertically is the “Piggy board” that provides the M860 with 8 digital input/output ports, and 4 analog inputs as well as 4 analog outputs. On the right, it shows the rear side connectors of an M860 without the openings for a Piggy board.
The “Piggy” expansion card, from left to right and bottom to top, adds 8 digital input/output ports (Micro-D D-Sub connector) and 4 analog outputs (numbered from right to left as 1 to 4), as well as 4 analog outputs (round connectors).

The difference between these two photos is the same type of M860 with 64MB RAM (middle board) and the analog (4in/4out) and digital (8) audio I/O board, also known as the “Pigggy Board” (bottom board).

When this board is referred to in technical literature, it is often under one of these acronyms: S.P.W. (Signal Processing Workstation), I.S.P.W. (IRCAM Signal Processing Workstation) and S.I.M. (Station d’Informatique Musicale), but these are misleading. In fact, the correct name is probably the M860 card. It’s important to understand that these acronyms actually represent the name of the complete solution, i.e. a NeXTComputer (cube) and one or more (maximum 3) M860 cards, as well as other peripherals (MIDI interface, audio interface for DSP, etc.), and the IRCAM software (Max/FST, SpecDraw, Spat, Circle, Animal, etc.) required to operate the whole!

IRCAM’s signal processing station, the following ports are connected: audio input/output ports, power supply, black-and-white monitor, RJ45 network, and SCSI.
8 versions of M860 board, 2 NeXT Cube CPU 68040 board, 4 Piggy board
Screenshot of a SIM where Max/FTS is launched; on the right are application icons for several versions of Max/FTS, including Animal, Signal Editor, and Circle.

For years, until the closure of IMEB 10 in Bourges, my wife Myriam and I presented compositions at the international festival “Synthรจse.” Below is the patch of one of my wife’s compositions, followed by the link to the player to listen to it.

The Max/FTS patch of the piece “Vagues ร  l’รขme”

Philippe Manoury – En รฉcho

While exploring the machines I recovered from IRCAM, I found directories belonging to famous users, including Philippe Manoury.

Here are two examples of Max/FTS patches (they are actually complex with multiple layers of patches) and videos that give an idea of what they result in.

Pluton

En รฉcho


IRCAM Signal Processing Workstation
Two of my S.I.M. in full operation, notice the use of the ports from the Piggy Board, ProPort 656, and RS423-MIDI.

Each board is equipped with two 40 MHz Intel i860 processors, 64MB RAM, and a 27 MHz Motorola 56001 DSP.

To find out more about the M860, read this article.

Intel i860 CPU
Intel i860 CPU

Ariel – Quint Processor (7000$)

Ariel / CCRMA Quint-DSP
Ariel / CCRMA Quint-DSP

The QuintProcessor is equipped with five Motorola 56001 DSPs at 27 MHz and a maximum of 16 MB, each with an external communication port. There’s also a built-in SCSI interface for use by the DSPs.

This board is a competitor to the IRCAM board. The Quint Processor was developed at the request of the CCRMA at Stanford University in the USA.

To find out more, click here : Real Time Sound Processing & Synthesis on Multiple DSPs Using the Music Kit and the Ariel QuintProcessor

The Ariel QuintProcessor [Ariel, 1990] is a board for the NeXT cube that contains five 27 MHz DSP56001 signal processing chips, each with its own bank of static RAM and pair of serial ports. The DSPs are arranged in a star configuration, with one โ€œhubโ€ and four โ€œsatellites.โ€ The 56001 is well-known as a low-cost and powerful signal processor that is well-suited to musical uses. The QuintProcessor (โ€œQPโ€) augments the power of the 56001 by providing the following additional capabilities:

  • 0 wait-state static RAM (32K words for each of the satellite DSPs and 8K words for the hub DSP).
  • 256K, 1M or 4M words of dynamic RAM for the hub DSP. Automatic refresh hardware for the DRAM.
  • Interprocessor communication hardware.
  • Two NeXT-compatible DSP ports and a larger connector that brings out six more serial ports.
  • SCSI controller and real-time clock for hub DSP.
  • Rapid NeXTbus access to the host interfaces of the DSPs and to the other QP hardware.

MIDI interface

On NeXT, MIDI events can be managed. The NeXTcube and NeXTstation motherboards feature two RS423 serial ports (with mini din 8-pin connectors). These serial ports interface directly with MIDI serial peripherals designed for older Apple Macintoshes (those with a serial modem/printer port).
These interfaces require no special drivers, as they are handled directly by the system!

Apple MIDI Interface Kit

Original Apple MIDI interface for Macintosh models with 8-pin mini DIN serial ports. The kit comes with three cables: a serial cable for connecting the interface to the Mac, and two MIDI cables for connecting MIDI musical instruments.

QUEST INC – MIDI LINK CONVERTORS (95$ a 160$)

midiman – MiniMacman 1 in/ 1 out)

midiman – Macman (1 in/ 3 out)

RCN – MIDI NeXT (1 in/ 2 out/ 1 thru)

RCN - MIDI NeXT avant
RCN – MIDI NeXT avant
RCN - MIDI NeXT arriรจre
RCN – MIDI NeXT arriรจre

Hardware is good, but software is essential !

Very quickly, there were many developments dedicated to audio, both for software that utilized the native audio capabilities of the Cubes and Stations, and especially for DSP acquisition boxes and later for the M860 and Quint cards.

Software provided by NeXT

This software is available in the standard installations of NeXTStep.

MonsterScope

The famous MonsterScope, software that functions as both an oscilloscope and a spectrum analyzer, uses the DSP 56K of the motherboard. It handles analog inputs from the microphone as well as signals from the DSP.

NeXT Monster Scope
Main window

Sound

The basic audio editor included with NeXTStep allows you to play and record audio through the microphone, as well as perform copy and paste operations with insertion.

There are other audio software programs, such as CDPlayerโ€ฆ


Research Center Software

Spectro3 allows for the analysis of the sound spectrum of an audio file.

Z-quencer

A MIDI sequencer.


Commercial software

Digital Audiometer

Digital Audiometer was designed to test hearing. The quality of the test signals was much higher than what was normally available at the time.

SoundWork

A software for managing sound acquisition from audio sources via DSP ports includes options like Digital Ears and Digital Microphone.

NoteAbility

A music notation editor / MIDI sequencer.

ModPlayer

A software that handles Mod files, which is a music format that is not very popular nowadays but was trendy in the early 90s.

Here he plays Close to the Edge by Art Of Noise!

Digital Audiometer

An audiometer, or acoumeter, is a device used in audiometry to measure an individual’s hearing capabilities. It provides a report on auditory acuity, indicating tendencies toward deafness or hearing impairment.

Sequence

A relatively powerful sequencer that handles both MIDI and audio.

Studio3 Setup

A software that allows configuring Opcode Studio3 interfaces.


Les logiciels de l’IRCAM

Max / FTS

Max/FTS is a software invented and developed by Miller Puckette in the mid-1980s at IRCAM, initially on Macintosh and ported to NeXT in 1990 specifically for Cubes equipped with M860 cards.

It manages MIDI and audio, and can operate without DSP cards, but it performs optimally when supported by M860 cards. Up to three of these cards can be installed in a Cube.

Please note, Max relies on the CPOS (Co-Processor Operating System11), which is specific to M860 cards and is loaded when the RESTART button is pressed in the FTS window, after selecting the cards and number of CPUs to use.

I have several versions of Max/FTS, ranging from 0.17 to 1.3.22, but version 0.26 provides the best balance of workload and efficiency.

The window allows you to activate the M860 cards and define the number of CPUs to use. Here there is a single M860 card positioned in slot 4, thus using 2 CPUs.
Here, there are three M860 cards positioned in slots 4, 2, and 6, thus providing 6 CPUs.
This window allows you to modify the settings of each M860 card and the Piggy boards, and this can be done hot-swappable, meaning without restarting. There’s always one M860 (in slot 4) configured to operate at 32KHz and it’s equipped with a Piggy board.
Here, three M860 cards (in slots 4, 2, and 6) are configured to operate at 32KHz, and all of them are equipped with a Piggy board.

Max has two modes: an execution mode and an editing mode. You switch between them by clicking on the key icon located on the left side of the patch window (highlighted in red here). You can also observe whether the CPUs are activated (if there are active CPUs, the red overlay with a cross is not present). I’ve added the CPU numbers in red. If a checkbox is grayed out, it means that the patch in the window is assigned to that CPU.

mode execution

When the editing mode is active, new icons appear.

mode รฉdition
6 CPU
4 CPU
2 CPU

Max/FTS is a graphical programming environment based on modular boxes and connections that allow for creating complex processes.

The modular boxes represent basic building blocks such as arithmetic, boolean, and mathematical operators, as well as data tables, controllers, buttons, and various input/output interfaces (analog, digital, MIDI). It also provides display tools like text, graphics, and matrices. This system is extremely powerful and flexible, enabling intuitive design and manipulation of audio data flows.

Max/FTS stands out for its ability to integrate and synchronize various types of data and media, facilitating the creation of complex interactive applications without the need for traditional textual programming. Since those early days, the Max user community continues to grow, contributing to a vast and diverse library of extensions and examples, further solidifying its position as an indispensable tool for creators and researchers.

A typical patch in Max/FTS

A small compilation of patch Max :

Animal

Animal12 (Animated Language) is a software package specific to the M860 board and dedicated to the rapid development of projects involving real-time signal processing.

animal patch

Spat

Spat is a software layer that spatialises sound, and is still being developed.

Circ

This is spatialization software that allows you to record a sound trajectory but only on the 2D plane (inside the disc shown in the following figure) for Spat.

The Circ application and a Max/FTS patch.

Gรฉnรฉrateur de trajectoires

It’s a spatialisation software that allows you to record a 3D sound trajectory for Spat.

Signal Editor

Audio editor13 from IRCAM, capable of doing the same thing as the NeXT editor, but it takes large files into account and allows you to enlarge the visualisation, thus offering greater finesse in copy/paste.


Les logiciels du CCRMA

Music Kit et DSP Tools

Music kit is a suite of functions provided by NeXT in NeXTSTEP that allows for managing all MIDI, DSP, etc. events. It is widely used by developers at CCRMA because this software library handles DSP 56K very well. Furthermore, the CCRMA’s Quint is based on 5 DSP 56K.


InstrumentBuilder is a program that makes it easy to prepare for using cmusic.


SynthBuilder is a software package that lets you build musical instruments and audio processing in the style of Max. It’s based on the MusicKit software library (and in particular the 56K DSP).


Vocal Trac Editor can be used for voice synthesis.


Some links

About development tools : MusicKit and SndKit Concepts

Links to Motorola DSPs and Intel CPUs: on the forum DSP NeXT. The I860s seemed to be difficult to program for maximum power… Read the next article : Floating-point performance of the i860.
I recommend the very interesting i860 64-Bit Microprocessor – THE ADVANCE INFORMATION 1989

From what I’ve read on the Benchmark, a NeXT 68040 at 25 MHz had an index of 116, while an i860 at 40 had an index of 26. That’s ~4.5 x faster!

Finally, here’s a list from a catalog NeXT summer 1992. It contains references to the following audio peripherals for NeXT:

DATA ACQUISITION AND SIGNAL PROCESSING
55 ADA1800 Digital Audio Interface Stealth Technologies, Inc.
55 A/D64x Singular Solutions
56 Ariel DM-N Digital Microphone Ariel Corporation
56 Ariel/IRCAM Signal Processing Workstation Ariel Corporation
57 Ariel ProPort Model 656 Ariel Corporation
57 Ariel QuintProcessor Ariel Corporation
58 DataDisplay Dazzl
58 Dazzl Analog-to-Digital Convertors Dazzl
59 Digital Ears Metaresearch, Inc.
59 Midi Link Convertors Quest Inc.
60 SCSI488/N IOtech Inc.
60 SoundHouse MIDIapolis Systems

Products Available Soon
62 Ariel DatPort Ariel Corporation
62 LogicStream Lab Interface Board And Virtual Instrument Toolkit LogicStream
63 Model MZ-4 Four-channel Analog- to-Digital Converter Greeneridge Sciences Inc.
Applied Speech Technologies AST A/D16

Thanks to : Andreas, David Fischbach, IRCAM.

Update on :

  1. Le fichier PDF de Science et Vie Micro (SVM) – numรฉro double 63 de juillet-aoรปt 1989 โ†ฉ๏ธŽ
  2. Le fichier PDF des Tarifs NeXT Printemps 1991 โ†ฉ๏ธŽ
  3. La brochure PDF du Pixar Image Computer โ†ฉ๏ธŽ
  4. Le fichier PDF NeXT 0.9-1.0 Release Description โ†ฉ๏ธŽ
  5. Le fichier PDF NeXTSTEP 2.0-Release-Notes โ†ฉ๏ธŽ
  6. La brochure PDF du Cube โ†ฉ๏ธŽ
  7. La brochure PDF du NeXTstations โ†ฉ๏ธŽ
  8. La brochure PDF du NeXTstation Color et NeXTstation Turbo Color โ†ฉ๏ธŽ
  9. La brochure du NeXT-External-CD-ROM-Drive โ†ฉ๏ธŽ
  10. Les publications de l’IMEB de 1970 ร  2010 โ†ฉ๏ธŽ
  11. Un article sur CPOS โ†ฉ๏ธŽ
  12. Animal: Graphical Data Definition and Manipulation in Real Time โ†ฉ๏ธŽ
  13. Un article sur Signal Editor โ†ฉ๏ธŽ

Flash Kit for Fairlight CMI

The FLASH KIT is the solution to do without the original 8โ€™โ€™ drive !

 

KIT Flash pour Fairlight C.M.I.
My CMI IIx

Why should you replace one of the 8โ€™โ€™ floppy drives of your Fairlight Computer Musical Instrument by this diskdrive emulator ?

There are three reasons :
Itโ€™s currently very difficult to find or to buy functional blank 8โ€™โ€™ floppy disks.
The floppy drives are often faulty or grimey, and it is more difficult to find a cleaning floppy or alignment floppy ! And what about a new functional diskdrive ?
The original floppy disks, which are often more than 30 years old, are very often unreadable because they are worn, damaged (by scratch, mildew) or corrupt (by magnetism).

The Fairlight C.M.I.โ€™s FLASH KIT is the solution to these problems and new features are added !

Indeed, the flash kit allows you to do the following :
save your sounds on your computer (Mac or PC),
easily exchange your sounds and compositions via Internet,
access to hundreds of virtual disks with out having to carry large and heavy 8โ€™โ€™ disks,
not have reliability problems disks that do not work at a crucial moment.

The Flash Kit includes :
documentation (Full part of the one that you are reading now) in English and French,
a SD type flash Memory card where the disk images are stored (delivered with more than 256 disk image),
a Flash drive (The black metal structure with many electronics boards),
a fifty way cable,
a thirty-four way cable with converter attached,
a two way power cable.

 

Items Kit
Full part (2013)

 

Full part (2014-…)

 

How does the flash drive work ?

The Flash player behaves like a 8″ floppy drive ! When you start and depending on the setting of the switches CMI nothing will happen as long as you do not select a startup disk image !

You can choose between several modes of operation the drive 8″ remaining, which by default is configured as drive 0 (boot drive) and flash drive that comes configured drive 1 (data drive).

The Flash player is able to emulate two virtual drives! The A and B ! The hole in the front is to insert an SD card.

The use of the flash drive is very easy. As it is based on the HxC drive, it works like it ! There are also three LED : green, orange and red.

The left and right black buttons allow you to scroll through the choices. The validation button is to confirm your choice. In order to have the configuration mode, press and hold the validation button for many seconds. A specific menu will appear on the LCD display.

The display contrast can be adjusted using the knob to the right of the front panel.

You have also a sound beep when the flash drive change of virtual head and when you move in folderโ€™s, fileโ€™s and when you confirm a choice.

Compatible flash cards are SDHC Card and up to 32GB capacity. They are being formatted in FAT32 format. Caution !! I do not recommend to use a Macintosh, because Mac OS/X creates hidden files that can affect the proper functioning of the Flash Player. Use a PC.

You can get more details on setting up the Flash kit by reading the documentation.

Documentation :

The last update of the installation documentation is in the file : FLASH-KIT UK R1a

Import WAVE files in a Fairlight CMII, II, IIx and III. With the flash cardย it is possible to transfer wave files between the Fairlight CMI and computers.
This small note is valid for Mac and PC. Click CMI Flash Kit tips and tricks to download the document.

But the best way to get an idea is to view some demonstrations on the following videos :

Some videos of Flash-kit customers:

 

Some pictures

 

The 1st big batch

 

The current version

 

 

Order Kit

You can order the kit by paypal or by bank transfer. I can also undertake to set up the Flash drive in your C.M.I., please contact me for more information. For half size floppy format for CMI III, thank you to specify that in the paypal message.

The total price of the Flash Kit is 500 euros, select the model and valid.

Fairlight CMI Kit
Fairlight CMI Kit
Flash Kit for Fairlight CMI I, II, IIx on left or for CMI III on right
โ‚ฌ500.00
Size :

There is a maximum time of one month between order and delivery. The kits are manufactured on demand.

The price includes delivery to all countries (with 500 euros insurance, tracking number, proof of shipping, Retail Delivery against signature).

For more informations on the transport www.colissimo.fr