A research team in the Computer Science department at Bradford University have developed a digital musical sound simulator, controlled by microprocessors, which combines flexibility and quality with moderate cost. It is able to reproduce a wide range of conventional instruments and can also be used to synthesise new sounds, as well as offering some unique features.

The work started over seven years ago when Dr Peter Comerford — a Senior Lecturer in Computing and a church organist — started experimenting with sound generation using a purpose-built computer. A microprocessor-based design soon followed; the NRDC spotted its commercial potential and provided financial support. A small research team came together under the direction of Dr Comerford, and a number of prototypes have been built.

The system is based on microprocessors; they are used for keyboard scanning (to monitor the player's actions) but also to control the sound production itself. The design uses mass-produced components: specialised hardware (to meet realtime demands) is kept to a minimum so the majority of functions can be manipulated directly by software, which makes the instrument both powerful and versatile. The programs are written in a modified version of Z80 assembly language.

The simulator is of general purpose application: the sounds it will produce depend upon the numbers which are loaded into memory. Therefore, you can either analyse recordings of conventional instruments and use this information to recreate the original sound; or you can create new and experimental sounds by manipulating the data in the simulator, using an interactive program; or with a suitable interface, the simulator could become a high-quality, low-cost digital synthesiser.

For simulations of conventional instruments, the team make tape recordings which are played into a PDP11-20 at the University. Using analysis programs written specially for the project, the sounds are broken down into their constituent partial tones. This data is stored inside the simulator and is used by the microprocessors for generating the correct waveforms. Each waveform can be very precisely defined, which is important in the accurate reproduction of sound. But there is more to realistic sound than tonal quality. Pitch, attack, decay, enveloping and so on must all be taken into account. For instance, the first application of the simulator was as an organ, and the team have found that many complex and interacting features must be incorporated if such a simulation is to be musically satisfying. For instance, the harmonic spectrum of each stop is not a fixed feature — low notes are generally richer in harmonics than high notes — so the simulator must produce waveforms which change in composition across the keyboard. A pipe organ can contain hundreds of pipes, many having complicated transient tones at the beginning of each note; the instability of 'pipe against pipe' is another factor which must be part of a satisfactory simulation. It is also essential to use an adequate number of independent sound generators, to create a 'large' sound instead of just a 'loud' one.

By loading the simulator memory with different data from mini-floppy discs, the sounds produced can be changed completely within seconds. Therefore, the range of available conventional instrument simulations is limited only by the effort required to analyse them. On the Bradford prototype, the Roman Romantic Church Organ can become a Continental Organ, with a completely different specification of more aggressive sound. The Bradford team have also produced a unit cinema organ, which not only has the distinctive voicings and tremulants, but also a range of instruments (each with its own complicated characteristics). Percussion includes tubular bells, glockenspiel, xylophone, chrysoglott, bass drum, cymbal and triangle; the partial tones of these instruments are not harmonically related, so each frequency has to be individually defined. Other instruments are a harpsichord with 'plucked tone', an orchestral trumpet, and a 'touch-sensitive piano', which is also able to simulate the effects of both multiple stringing and sharpened tuning.

One of the Bradford prototypes has been on trial 'in the field' at a local church, where the specification swapping facility has made it useful for a wide range of music.

Using a video display terminal and a specially written interactive 'voicing' program, it is posible to make direct changes to the data in memory — and to store the changes on mini floppy disc — so that the parameters defining each sound can be modified in seconds while you listen. Thus an organ tone can be quickly and easily altered to suit the acoustics of a particular building or the taste of a particular player. (This facility can also offer pipe organ builders the opportunity to experiment with adventurous voicings, in a way which would be prohibitively expensive using pipework.)

Besides a new concept of sound generation, the Bradford system offers a number of special 'features'. For the player who is too nervous to play well in public, there is the chance to pre-record a piece and then play it back indistinguishable from the original performance. The floppy disc is used to record the music as a sequence of events — notes pressed, stops selected — a principle reminiscent of the pianola. The microprocessors then 're-enact' each recorded event, so that the piece is reproduced exactly as it was played. The playback speed can be altered without changing the pitch of the sound. Transposition is a matter of pressing a button and the instrument can also be 'fine tuned' in seconds to match the pitch of, say, a pipe organ or a piano, so that duets can be played.

Research continues at Bradford, to refine voicing and produce more simulations, as well as in other areas.