You are here

Typhoon OS, Yamaha TX16W

Alternative Operating System For Sampler
Published June 1994

The TX16W was about to go down in the annals of electronic instrument history as a brilliant‑sounding sampler handicapped by a pig of an operating system — until a group of Swedish users took matters into their own hands. Duncan Werner investigates the ultimate in third‑party development...

Typhoon is an entirely new operating system for the Yamaha TX16W sampler. It was created and is being further developed by NuEdge Development, a group of Swedish TX16W owners who decided to attempt the programming of an operating system that utilises the instrument's potential more fully than the original Yamaha operating system.

About The TX16W

The TX16W was first released mid‑1988 and had a retail price of around £2000. It was not a popular machine, due to its long‑winded operating system; I remember talking to one Yamaha R&D person around this time who described the TX16W as "over‑engineered". What he meant was that the TX16W offered a plethora of functions, but they were hidden within a very user‑unfriendly operating system, often described as slow, complex and cumbersome. When Yamaha realised it wasn't really working out, production ceased and dealers sold the remaining machines off at under £1000.

Fortunately, however, the TX16W's operating system was stored on disk and was designed to be loaded into the machine upon booting up and consequently into RAM. Future upgrades could simply be installed by loading the appropriate software upon booting. This is where Typhoon comes to the rescue, five years later.

The TX16W is a 12‑bit linear conversion AWM (Advanced Wave Memory) sampler, with 16‑voice multitimbrality and 16‑note polyphony, and a total of 10 separate audio outputs. It was originally marketed by Yamaha as a Digital Wave Filtering Sampler; one of its selling point was a comprehensive range of dynamically variable filters. However, with the standard 1.5Mb of memory, the Yamaha OS and filters took up so much room that there was hardly any memory left for sample data. An expansion memory board was required to load anything more than just a few waves. By contrast, the entire Typhoon operating system, including filter tables, is loaded in less than 30 seconds, using 500Kb less memory than the original operating system, effectively giving back to the user one third of the 1.5 Mbytes of standard memory. The fixed internal memory of 1.5Mb can be expanded to 6Mb by inserting three further memory expansion boards.

As NuEdge Development are TX16W users themselves, they are open to ideas and suggestions as to the future features of Typhoon. Users can thus give feedback directly to the programmers and possibly influence Typhoon's future design. NuEdge are keen to point out that they will remain committed to the product for a long time to come, and stress that version 1.0 is not intended to be the main release version of Typhoon. They hope to be releasing version 1.1 within a couple of months, and this will be free to existing version 1.0 users.

Typhoon Sound Architecture

Typhoon's sound architecture is built upon three categories known as Items. These are Waves, Voices and Performances. Each category can be considered to be a list within which Items are stored (see Figure 1).

So, from right to left:

  • Waves represent single 'pieces' of sound, such as a drum or piano key
  • Voices use Waves in their construction, to represent an instrument (drum kit or piano, for example).
  • Performances use a collection of Voices, which can be played simultaneously via MIDI.

There is no restriction on how many Items may be used by any other Item. For example, a Performance Item may consist of as few or as many Voice Items as is required; similarly, a Voice Item may use any or all of the Wave Items available. So an Item may be intensively shared, forming a complex Item hierarchy (see Figure 1.2).

Sample Display

The sample screen allows the user to select which functions will be automated during sampling. These are:

  • Trim: automatic editing of the front and end sections of the wave.
  • Info: performs an analysis on the sample data and provides information about its dynamic range, which is given as a percentage. The number of clipped samples is also calculated when Info is selected, allowing resampling if the user considers that too much (or not enough!) clipping has taken place.
  • Normalisation: maximises the Wave sample levels such that the peak levels reach the maximum quantisation level (all 12 bits on).
  • Pitch: provides automatic tracking of the Wave's pitch.

After the automatic settings have been chosen, returning to Sample on TX16W's front panel gives three further fields:

  • External Trigger: via MIDI or footswitch which plugs into the front panel.
  • Sampling Frequency: mono at 16, 25, 33 and 50kHz, plus stereo sampling at 33kHz.
  • Sampling Time: this defaults to the maximum sampling time available for a given sample frequency, from 5.2 seconds at 50kHz up to nearly 16 seconds at 16kHz, with stereo sampling up to 8 seconds at 33kHz. The sampling time is variable up to these hardware limitations. Typhoon treats and edits a stereo waveform as a single Wave, so stereo looping is performed without needing to loop independent Waves — though the Waves to be looped must be of the same length. A stereo Wave can be sampled directly from the stereo input on the front panel of TX16W or it can be created by joining any two mono Waves together in Wave edit. More about Wave edit later.

Pressing Go now takes you to the final sample window, where the sample Threshold level can be user defined or automatically set. The auto setting pre‑samples the Wave by 250ms before it reaches the sampling threshold, to prevent loss of transients. Once a Wave has been sampled, it can be auditioned, then you can either 'keep' the Wave or 'retry' sampling. When a wave is kept, it is automatically entered into the list of Entries within Wave Edit.

Wave Edit

As with the Yamaha OS, Waves are the most fundamental item in Typhoon, and are the lowest level of the architecture hierarchy. Various functions are available within Wave Edit, including Trim, to manually or automatically set start and end points of a Wave; Loop, to set the 'from' and 'to' points within a Wave, manually or automatically; Crossfade, for looping smoothly; and Join or split, which joins two mono Waves to form a stereo Wave or splits a stereo Wave into two mono Waves

Also available are Sample rate conversion, for resampling at lower sampling rates; Gain adjustment; Normalisation; and automatic pitch tracking.

Associated with Wave Edit is the Wave Edit Utilities menu. There are seven Utilities dedicated to Wave Edit: Load — Typhoon will load in three different Wave formats: Compressed (this loads Waves stored in Typhoon's default format), AIFF (Audio Interchangeable File Format), and Yamaha (original Yamaha Wave format); Save (same formats); Delete; Rename; Copy; Swap, which allows two Waves to exchange positions; and Dump, via SDS to the MIDI port, or to the RS422 port.

Voices And Groups

In Typhoon, Voices represent instruments. In order to provide the necessary features a Voice may require, a large number of parameters are available. A Voice is partitioned into Groups of parameters, and The Voice Edit page contains the Group settings for each Voice.

Parameters adjustable in Voice Edit include Range — both the key range of the group and the velocity range which the Group will respond to; Pitch, to pitch‑shift the Waves used by the group, in octaves, semitones and cents; Volume, which specifies the volume and velocity‑sensitivity parameters of the Group; Filter, with 16 filter types available (see Figure 4); Output; and Mode, which specifies the playing mode of the Group. Four modes are available: Normal; Oneshot (useful for percussion); Glide (monophonic portamento); and Release (triggers on key release).

The Voice Edit page also hosts a comprehensive modulation map. There can be up to eight different modulations taking place in real time, with 15 modulation sources and 13 destinations (see Figures 5a & 5b). Also available are two low‑frequency oscillators, each independently configurable, with five available shapes — triangle, sawtooth, square, sine and noise. Two independent, 3‑stage envelope generators are also provided. These are general envelopes which can be applied in the modulation table within Voice Edit.

Associated with Voice Edit is the Voice Edit Utilities menu — offering standard functions like Load, which loads from disk the selected Voice with its respective Waves; Save, which saves the chosen Voice as a Voice file to disk; Delete — only Voices are deleted; Waves remain, as they may be shared by other Voices; Rename; Copy; Swap; and New, which creates a new Voice with default parameters

Performance Edit

Each Performance consists of three main settings; Entries, program change codes, and external triggering options. An Entry can be considered a building block upon which a Performance is built (see Figure 6); each Entry has settings for Voice, Output, Transposition and Priority.

  • Voice: here a Voice is selected, assigned to a MIDI channel and given a volume setting. This style of building Performances supports both layering of several Voices and multiple use of a specific Voice on various MIDI channels (see Figure 4).
  • Output: a selection of any, left, right, mono, stereo, or individual outputs are available for each Entry. This would enable, say, 10 Entries, each with different Voices and MIDI channels, to be directed to any or all of the 10 separate audio outputs (eight Individual plus left and right). For stereo Entries, there is a pan parameter. The individual output selection overrides the output settings in Voice Edit.
  • Transposition: the Voice for an Entry can be transposed in octaves, semitones or cents.
  • Priorities: this is a major feature of Typhoon, providing dynamic voice allocation and overcoming one of the main difficulties with the Yamaha OS. It allows the user to define priority settings of low, medium or high for individual Entries in a performance. A channel playing a higher priority note will override one with a lower priority.

The remaining Performance settings are Program Change (to give MIDI control of selected Voices), and External Triggering (to allow sounds to be triggered from a footswitch which plugs into the TX16W's front panel).

As you'd expect, Performance Edit also has its share of Utilities — Load, Save, Delete (again, deletes Performance Items only; Voices and Waves are kept, as they may be shared with other Performances), Rename, Copy, Swap and New, for creating a new performance with default settings. Also here are System set up (where preferences for the whole system are defined), Master settings (volume for left and right outputs, master tuning and device number for SysEx dumps via SDS), External Controllers (two definable external controllers for modulation can be assigned controller numbers), and control of the Data Slider, which allows menu scrolling, value editing and similar changes to be controlled from an external MIDI source using a data entry slider. The Remote option allows controller numbers or keys to be set up for external control of front panel buttons on the TX16W keypad. Some keys are not implemented, however, so complete control is not possible.

All data transmissions that can be sent over MIDI (SDS) can also be sent out via the RS‑422 port on the rear of the TX16W. The RS‑422 option in Performance Edit Utilities allows you to set one of four data transfer rates: 19.2, 38.4, 76.8, and 153.6 Kbaud.

Experimental Sounds

Included with the system disk are some 'experimental' sounds, which demonstrate the capabilities of Typhoon. The drumkit sounds are interesting in that they have been constructed from very simple waves, such as sine and noise, then multi‑layered.

The setup includes five performances, namely:

Drumkit: a set of drums similar to the Roland TR808/909 sounds. Sine, Noise and Saw waves are modulated, filtered and layered extensively (not necessarily in that order!)

New Age: consists of the voices Fairlite, Analogue Strings, Ice Rain and Noise Wave.

Wavebell: a performance built upon a multisampled bell waveform.

Multi: a multi‑channel performance containing the voices Analogue Strings (Channel 1), Unison (Channel 2) and drumkit (Channel 10). Program change for voices is enabled.

16‑channel: contains 16 empty Entries, one for each MIDI channel. This is a useful starting place for your own setup.

Typhoon In Use

The Typhoon manual is basically very good. Four chapters cover Sound Architecture, Storage System, User Interface and a Reference section. A tutorial section would be a good idea for first‑time samplists, though.

After working through the process of recording a sound from scratch through to creating a multi‑channel performance, it became apparent that the reference section of the manual would be more intuitive if it started with basic sampling and worked through to creating voices and performances; the Version 1 manual seems to go about things in the opposite way. Hopefully, further versions will address this.

A very practical feature of Typhoon is its 'hear what you see' approach. Generally, as you move through the operating system and perform edits, the audio output mirrors the Typhoon parameter changes. This is common sense, really, and is a big improvement over the Yamaha OS. Gone are the edit buffers and the boring "Are you sure?" messages. Typhoon implements destructive editing with undo features, which provides quickly reversible editing of Items.

The Yamaha OS was not completely memory resident. This meant that to perform multiple edits you were often asked to "insert system disk". The entire Typhoon OS resides in memory, leaving the disk drive to deal with data files.

You no longer have to deal with assigning 'soft oscillators' to a Voice. The Priorities feature deals with dynamic voice allocation. If the polyphony is being eaten up by a particular Entry within a Performance, this is easily solved by giving this Entry a lower priority. The programming of the polyphony for a particular Voice was one of the more limiting aspects of the Yamaha system. Nice one NuEdge!

I used a Yamaha SY77 synth as the master keyboard to communicate with TX16W/Typhoon. There are many ways to configure the modulation table to give real‑time performance parameter changes. I tried a few, including assigning a modulation wheel to the attack parameter of the AEG and controlling the dynamic frequency axis of a high‑pass filter from keyboard aftertouch. In another setup I arranged for LFO1 to modulate the frequency axis of a high‑Q filter. Using chords, this produced an arpeggiated effect as the filter highlighted different parts of the spectrum. The potential here is almost limitless, considering the available modulation sources and destinations.

I also connected the TX16W to ye olde Atari STF to load in some waves created elsewhere and saved in Digidesign Sound Designer format. Communication between Typhoon and Sound Designer via MIDI SDS (Sample Dump Standard) was very straightforward. There are, and have been for some time, sample libraries available for the TX16W. But there's now talk of an international user group formed by registered Typhoon users, which could be a forum for the exchange of voices in Typhoon format.

By the time Typhoon becomes 'mature', the TX16W/Typhoon combination will be a very powerful sampling performance instrument. The only limitation that I could gripe over is the screen display of the TX16W. With space for only 40x2 characters, you constantly have to scroll the screen display horizontally to make parameter changes. An external monitor or remote control using something like the MIDI manager in Cubase would be the icing on the cake.


NuEdge claim that Typhoon transforms the TX16W: this is certainly true. The extensive modulation mapping features alone turn the TX16W into a very playable performance instrument. There really is no question about the comparison with the original Yamaha OS; if you have or use a TX16W, then Typhoon is a must.

Memory expansion boards are no longer available from Yamaha, but Sounds OK can supply them from a third‑party manufacturer to order at a cost of around £120 for a 1.5 Meg board. Thanks to Sounds OK for the loan of Metrasound studiosamples disk set for the TX16W.

Main Features & Improvements

  • Completely new user interface.
  • Dynamic Voice allocation.
  • Destructive editing with undo.
  • Object‑oriented managing of individual Items.
  • New voice architecture with groupable parameters.
  • Eight modulators per channel with 15 sources and 13 destinations.
  • Two independent LFOs per channel.
  • Two independent four stage envelopes per channel.
  • Monophonic portamento mode and several other new playing modes.
  • Stereo waves treated as single waveform.
  • Up to 250 performances, voices and waveforms in memory.
  • Pre‑triggered sampling threshold.
  • Automatic loop finder, pitch tracker and waveform trimmer.
  • Resample utility for any sampling rate between 3Hz and 50kHz.
  • Five times faster start‑up time.
  • Up to four times faster saving and loading.
  • Audio file compression.
  • Reads and writes AIFF format files.
  • Incremental saving allows only modified data to be saved.
  • Requires substantially less memory than the original Yamaha operating system.

The Shape Of Features To Come

A couple of extra features will be implemented in Typhoon Version 1.1:

  • Import routine: this will load and translate Yamaha setups (and their Performances, Voices and Waves) into the format used by Typhoon. Individual Yamaha Waves can already be loaded directly into version 1.0.
  • System backup routine: future upgrades will be copy protected; the registered user will be able to make a single backup copy of the system software.

More features are planned for the future, including a Mix utility, to allow two Waves to be mixed together; a Reverse Wave utility, already available in the Yamaha OS but not yet in Typhoon; Filter table edit, which will implement the Filter Edit keypad on the TX16W not implemented in Typhoon version 1 (this has been reserved for future use, as the developers are creating filter editing tables which they say will produce such effects as filters with very low cutoff frequency, distortion, phasing effects, and so on). DSP functions are also planned — just a few of the possible DSP functions which may be developed include time compression/expansion, compression/limiting, reverb, vocoding, and others! Interfacing to a PC via the TX16W's RS‑422 port is also amongst the list of future possibilities that NuEdge are considering. This would enable high speed communication for file serving to/from hard disk.

System Setup Utilities

  • LOAD: an entire setup, including Performances, Voices and Waves, may be loaded from disk.
  • SAVE: saves the entire status of the TX16W as a setup file on disk. Options are available to save only the files not currently saved.
  • MEMORY: displays the amount of free main, system and filter memory.
  • ITEMS: shows the number of Performances, Voices and Waves available in memory.
  • DISK: names, renames or formats disks. The amount of free memory available on the disk is also displayed.
  • FILE: this is a file managing utility displaying file names, type and size. File types are: Setup, Performances, Voices, Waves, Filter table, AIFF, Yamaha Waves and System files.
  • PERFORMANCE SELECT: the active Performance is selected here, giving an option of three displays: MIDI (indicates MIDI channels and activity); Audio (indicates audio output activity); CPU (indicates the load on the TX16W's CPU).


Anyone using TX16W/Typhoon and who would be interested in a UK user group can contact Duncan Werner at the Faculty of Science and Technology, University of Derby, Kedleston Road, Derby, DE22 1GB. Fax: 0773 602601.

The NuEdge Development programm‑ing team can be contacted at: NuEdge Development, Fredrika Bremers g.24 S‑129 50 Hagersten, SWEDEN. Fax: 46‑8‑889427.