ÐÓ°ÉÔ­´´

WHEN the pop musician Stevie Wonder gave a concert in London last year,
his band of backing musicians suddenly stood up and walked off stage, leaving
many of their instruments still playing. At first the audience thought they
were hearing a tape recording. But then Wonder stopped singing, talked to
the audience, and then started again, with the electronic musical instruments
stopping and starting in perfect time with whatever he played on the keyboard
of his electronic piano. No tape recorder is fast enough or accurate enough
to perform this trick.

This year Leonard Bernstein, the conductor, also surprised his audience
at a concert in New York’s Lincoln Center when he stopped his orchestra
in its tracks and pressed a key that instantly replayed the last few minutes
of the performance. There had been no time to rewind a tape.

The technology that made both these stage tricks possible is called
tapeless recording. It relies on hardware originally designed for the computer
industry and now modified to record sound. It gives studio engineers and
performers unprecedented control over their music, allowing them fast and
accurate access to any part of a recording. They can also manipulate sounds
just as one might manipulate text in a word processor. The technology is
maturing so quickly that it is already impossible to tell when it is being
used simply by listening to the end product.

In a conventional tape recorder, sound is stored either as an analogue
waveform or as digital code on a length of tape. Inevitably it takes time
to search out a sequence of sound, by shuttling the tape backwards and forwards.
Studio engineers and musicians have always dreamed of an alternative; up
to 30 per cent of the time they spend in a studio may be wasted waiting
for tape to rewind as they search for a passage of music.

In a computer, digital code representing data or text is stored temporarily
in solid-state memory known as random access memory. As the name suggests,
such RAM allows computer operators to seek out their data and process them
with virtually no delay. This digital code, instead of representing text,
can be made to represent sound, with a quality comparable with that of a
compact disc when reproduced.

But RAM is also volatile; the digital code disappears for ever when
the power is switched off. Back-up power sources can guard against losing
data, but this is an expensive option. For this reason computers are designed
so that they can also record or ‘save’ code from memory onto a magnetic
disc. This disc can be a ‘floppy’ disc, so called because it contains a
thin sheet of plastic material, like tape, or a ‘hard’ disc, made from a
rigid metal platter. A computer operator has to load floppies into the computer
and remove them after use; hard discs are permanently fixed inside the computer,
and can store more data than floppies: tens of megabytes compared with around
1 megabyte, which is equivalent to 8 million bits of information. This is
important because sound takes up far more room in memory than text.

Inevitably, there are also drawbacks with using hard discs to store
data. The magnetic head, which records (writes) and replays (reads) the
data, is slow at retrieving code compared with the speed with which RAM
finds a selected sequence. Also, computers read code from disc in short
bursts, which is fine for text, but less useful for reproducing continuous
sound. Tapeless systems get round this by reading the code from disc via
a solid-state memory that acts as a buffer. This buffer converts the discrete
bursts of data into a continuous stream.

Tapeless recording systems therefore need to use a carefully balanced
combination of both magnetic disc and solidstate RAM. The balance exploits
the most useful features of each; the speed of access and quality of reproduction
of RAM, and the vast storage capacity and relative cheapness of magnetic
discs.

There are three broad areas in which tapeless technologyis changing
the face of the music industry: recording, editingand live performances.
At their most basic, such tapelesssystems simply reproduce the stored sound
as a replica of the original, just like a conventional compact disc or tape
recording. Many small home studios and workshops now use tapeless systems
for recording directly to disc instead of to tape.

Once a sound is stored as digital code in solid-state memory or on disc,
engineers or musicians can process that sound by whatever software they
choose to run on the computer that controls their system. They can alter
the tonal character, pitch and tempo of the sound, or convert the stored
codes into a sequence of control signals that can trigger different sounds
or instruments – the approach adopted by Stevie Wonder in his show. All
this can be carried out from a touch-sensitive screen, a standard ‘qwerty’
keyboard or an electronic piano keyboard.

Tapeless systems are useful in patching up a recording of an artist’s
performance. Vocalists, even trained professionals, may find it hard to
hit one note in a song with perfect accuracy, especially if a high note
follows a series of low notes. Slight imperfections that pass unnoticed,
or are soon forgotten, at a live concert can annoy listeners when captured
on a recording. For this reason it is standard practice in recording studios
for a vocalist to record a song and then replace any imperfect notes by
the process known as ‘overdubbing’. The engineer rewinds the tape a few
seconds before the mistake, and the vocalist listens on headphones. As the
imperfection comes up, the engineer switches from ‘playback’ to ‘record’
and the singers repeats the note. This is tricky to do, and the engineer
must often rewind the tape to play it back many times before the singer
hits the right note. Even if only a few seconds are lost rewinding a short
length of tape it is enough to rob the singer of spontaneity. A disc recorder
can retrieve the selected passage with almost an imperceptible delay, allowing
the engineer to replay a musical phrase over and over again, in a loop of
sound, with the singer trying each time round to achieve perfection. A further
advantage of tapeless recording is that when pop musicians overdub on tape,
by playing the same passages over and over again, the tape physically degrades.
Analogue tape recordings lose high frequencies, so cymbals start to sound
dull, while digital tapes accumulate errors. If the music is first recorded
onto a magnetic disc, before overdubbing begins, there will be no degradation.

Tapeless systems have other advantages over tape in editing sound. Television
and film studios often use them for editing recordings originally made on
tape and then transferred to disc. Editing tape means cutting and joining
with sticky tape. On a disc, the sound is spliced electronically and if
this does not achieve the desired effect, the editing engineer need only
press the button marked ‘undo’ and start again. Producers can change the
tempo of recordings without altering the pitch, or vice versa, simply by
lengthening or shortening the digital code representing a sound. This can
be useful when trying to match recordings of music, speech and special effects
recordings to pictures on screen; with tape, altering the speed of a recording
alters its pitch as well.

With the disc system, it is easy to mimic the ‘global replace’ function
offered by word processing, which searches text for every occurrence of
a chosen word, such as ‘teeth’, and replaces it with another word, for example
‘tooth’. In a disc-based recording system, the control computer can search
for every sound of a gunshot and replace it with a louder or quieter effect.
With conventional tape, the engineer can alter these sounds only by searching
for each one and processing it individually.

The recording engineer also has the chance to clone the same sound and
record it many times on top of itself. If this is done with a very slight
change of tone or timing, the result is a sound that gets fatter and fatter,
not just louder and louder. After repeated cloning, one voice can sound
like a choir, one helicopter can sound like a squadron.

Tapeless systems have already opened up new ways of linking visual performance
with musical embellishments. In the film Who Framed Roger Rabbit, there
is a sequence where two cartoon ducks play a duet on real pianos in a night
club. The duet turns into a duel and the tempo of the music accelerates,
yet the keys operate in time with the ducks’ fingers. Finally, the piano
lid opens to reveal a cannon. The music and synchronised sound effects were
recorded on a hard disc, along with a stream of inaudible digital pulses
that triggered solenoids to operate the piano keys. Synchronising sound
and pictures as the tempo gradually increases but the pitch remains the
same would be virtually impossible with a tape recorder.

The Disney theme park features live parades with cartoon characters
travelling on floats down the streets performing in time with recorded music
and sound effects. The organisers of the parade simply record the sounds
they need on disc, so that they can be triggered by pulses generated as
the floats pass reference points along their routes. Fireworks explode in
exact time with music, despite the distance travelled by the sound and variations
in its speed of propagation caused by changes in the weather. Music to accompany
the fireworks is also held on disc, and the fuses are lit by electric signals
whose timing is tailored in advance to suit the atmospheric conditions.

A British company called Autograph, which specialises in sound effects
for the stage, used a tapeless system called the Soundstation to create
the sound of surreal helicopters live on stage for the musical Metropolis,
and to fade betweenthe sound of trains and music for another popular musical,Aspects
of Love. Engineers backstage can synchronise prerecorded sound and music
effects with the actors even when they miss cues, simply by triggering the
sounds from a touch-sensitive screen at the right moment.

TVS, the commercial television station, uses a disc recorder for its
game and quiz shows. Events on the screen trigger a library of prerecorded
sound effects and musical jingles with great accuracy. Film and television
studios often need to replace poor quality dialogue recorded on location
with the same dialogue recorded in a studio weeks or months later. They
now use disc recorders to tailor the length and timing of the fresh dialogue
to fit an actor’s previous lip movements, with accuracy of one-hundredth
of a second. This is done by converting the original dialogue into digital
code. Although the original sound is of poor quality, it will match the
lip movements on screen perfectly because it was recorded as the actor spoke.
The actor can then read the same lines while watching a replay on screen,
synchronised to an accuracy of around a second. Software breaks the new
dialogue down into digital samples, compares them with the originals and
cuts, stretches and compresses the new samples until the old and new dialogue
match exactly.

The rock musician Pete Townshend, of The Who, has always had a keen
interest in the technology of recording. He has his own recording studio,
called Eel Pie Studios, on the banks of the River Thames near Richmond.
He has developed a novel technique for giving musicians the chance to add
or edit music onto pop videos and films, following the instructions of a
conductor, without any need for the conductor to be there in person. This
works by running a tapeless recording of the existing music in synchronisation
with a video-tape system.

The magnetic discs are electronically locked to four video-tape recorders
holding sequences of film that the new music is to accompany, together with
shots of Townshend conducting an imaginary orchestra. When the engineer
runs the magnetic discs and video recorders together, the musicians can
watch the film sequences and conductor on a bank of television monitors
and play in time with them. This way Townshend can be sure the pictures
and new sound will remain in perfect synchronism, however many attempts
are needed to produce the recording. Paradoxically, this robs the tapeless
system of one of its main advantages, avoiding wasting time while tape rewinds,
because the audio-disc system has to wait while the video tape rewinds.
The next logical step, therefore, is to bring together tapeless sound systems
and tapeless video systems, such as optical video discs read by lasers.

A byte on the town

It was inevitable that pop musicians who had used hard disc systems
to record albums at home or in studios would start to use the same technology
as part of their stage act, as a means of blending prerecorded sound with
a live performance. It is often impractical for a musician to carry a full
disc system on tour, so most store their samples of prerecorded music on
a floppy disc and load them into hardware hired locally. The library of
samples can be very large, so is often stored on optical discs 30 centimetres
wide, which can store up to 2 gigabytes of data.

New England Digital, the company that makes the Synclavier tapeless
system, is a pioneer in the field and the major supplier of tapeless systems
to musicians. Sting, another rock musician, recorded his album, Nothing
Like the Sun, on a Synclavier tapeless system in his New York apartment.
Recognising that most people do not want to go to a concert to see a computer
play, Sting hired musicians to go on tour with him and recreate most of
his music live each night. But the background vocal sounds were recorded
on disc and replayed under the control of signals generated electronically
from Sting’s guitar and piano. The singers had no complaints; they stayed
at home but were paid for the tour.

Stevie Wonder makes a stage joke of having his musicians stop playing
while he controls the reproduction of sound from a Synclavier. Some artists
are less honest. They have found that it is easier to achieve perfection
on stage if their musicians mime to disc recordings. The uninformed audience
has no way of knowing the truth.

The tapeless recording industry is now locked into a battle over the
storage capacity of the different systems available. Most can be expanded
to control four hard discs, known as Winchester discs, each with a capacity
of 320 megabytes; such systems can store nearly four hours of sound of the
quality of a compact disc recording in mono, or two hours in stereo. The
latest discs can each hold 512 megabytes of information, increasing recording
time by about 60 per cent. NED’s Synclavier system can now incorporate up
to 128 megabytes of RAM, equivalent to 200 personal computers, to give nearly
25 minutes of storage in mono.

However, because of the cost and heat generated by RAM, most systems
make do with around 32 megabytes or less. This is usually more than enough
because systems such as the Synclavier use a clever process to provide the
instant access to RAM that is vital in reproducing musical samples as part
of a live performance. The system holds a large collection of samples on
magnetic disc. The control computer then looks ahead at whatever sequence
of sounds the musician or sound-effects specialist has prepared, and pulls
each sequence off the disc and loads it into RAM just before it is needed
for reproduction. The sample is then deleted from RAM as more are pulled
from the disc – all the original samples remain safely stored on the disc.
The prices for tapeless studio hardware vary wildly, but anyone tempted
to dabble should be prepared to pay at least $100,000; for the latest and
most powerful equipment, prices can rise to $500,000.

As always, new technical advances bring fresh problems. Digital tape
is a remarkably robust medium, able to withstand bad treatment. If it is
broken, it can be easily joined again with very little loss of data. In
contrast, discs and RAM can suffer catastrophic loss. Anyone who has ever
used a computer knows that sooner or later, often for no discernible reason,
it will stop working in spectacular fashion. At worst, everything stored
in the computer’s memory can be lost, and all the data stored on the hard
discs corrupted.

Manufacturers of tapeless recording systems provide back-up systems,
which store on tape in a cartridge all the data held on both the solid-state
memories and discs. Backing up, and then checking the back-up tape by loading
the data back onto disc, can easily add half an hour to the end of an all-night
recording session. But as one pop music record producer put it: ‘Computer
recording has brought new disciplines. However tired or stoned you are,
you are never too tired or stoned to back up.’

Some perils are less obvious. When Stevie Wonder played at the concert
held at Wembley Stadium to celebrate Nelson Mandela’s 70th birthday, he
flew in from Los Angeles with a 30-centimetre optical disc holding all the
sound samples and sequences needed for his live performance. Just before
Wonder was due to go on stage the sound engineers went to load the Synclavier’s
control program from floppy disc – and found that someone had stolen the
disc drive that the floppy plugs into. So Wonder went on stage late, without
the musical programme he had prepared. He sang different songs and won the
day. A lesser musical talent would have been unable to perform.

* * *

Spinning a different kind of disc

HARD discs were christened ‘Winchesters’ because this was the code name
that IBM used for its project to develop the first rigid magnetic discs
to be fixed permanently inside computers. These discs are aluminium platters,
coated with a magnetic alloy of nickel cobalt. They rotate at 3600 revolutions
per minute and data streams on and off the disc at around 10 million bits
of information per second.

There is no contact between the head that reads the disc and the surface
of the platter – the head flies on a very thin cushion of air. This means
that neither the head nor the disc is subjected to wear so there is no loss
of quality, or build-up of digital errors, if the disc is played over and
over again during editing. The head can find any part of the disc in less
than a tenth of a second.

Usually 1 megabyte of solid-state RAM is used to buffer the 512-kilobyte
blocks of code read from the disc and to convert them into a continuous
flow of several million bits per second. This may extend the time it takes
to find and start playing a musical passage to a second or so.

Magnetic discs are ‘formatted’ before they are used. This means their
surfaces are mapped into labelled areas to make it easier for the computer
to search through them to identify and avoid any blemishes on the discs
that may corrupt a recording. With conventional tape systems there is no
way of knowing in advance whether there are faulty patches in the coating
of the tapes.

* * *

The ancestral heritage of the tapeless recorder

TAPELESS recording was born from both musical and computer parentage
in the early 1970s at Dartmouth College in New Hampshire. Sydney Alonso
and Cameron Jones were working at Dartmouth on computer-controlled synthesis
of sound. They developed a computer-based tool for teaching musical theory
and composition, and built a system that could synthesise music from digital
code. It recorded live sounds, broke them down into shortsamples of digital
code and played backsequences of samples to create newsounds based on the
original.

Alonso and Jones founded New England Digital in Vermont in 1976 and
started to make the Synclavier, a portable version of the Dartmouth system
that musicians could use on stage and in studios. Early models relied on
small banks of RAM to store temporary samples of sound loaded from floppy
disc.

In 1985, NED increased the storage capacity of the system by joining
the Synclavier to much larger quantities of RAM, and to Winchester discs
and optical discs read by lasers. The company kept most of the system under
the control of a 76-note piano keyboard because musicians generally feel
happier with this than with the keypad of a computer. This keyboard senses
not just the notes played but also the pressure, velocity and ‘touch’ of
the key stroke.

Studio engineers tried to use the Synclavier to record live sound directly
onto discs, so in 1987 NED offered them a16-track disc recorder. NED is
trying toregister the descriptive name ‘Tapeless Studio’ as a monopoly Trademark
and now has 600 Synclaviers in use around the world. Stevie Wonder was one
of the first artiststo adopt the system and Michael Jackson now has six.

Two British companies have developed systems that use Winchester discs
and RAM in much the same way as the Synclavier. These are designed primarily
for use by recording studios rather than by performing musicians. The Audiofile
disc recorder, from AMS in Burnley, Lancashire, is widely used by radio
and television companies around Europe as an editing tool and as a means
of storing large libraries of sound effects. And the Soundstation from Digital
Audio Recording, of Chessington in Surrey, is used by the BBC and Hollywood
film studios.

Although NED remains the leader in music computer technology, many Japanese
companies (notably Akai, Yamaha and Casio) have built cheaper systems that
can also store sound as digital samples.

* * *

A battle of bits to find time for musicians

THE computer that controls the Synclavier system from New England Digital
is capable of handling up to 16 separate channels of sound simultaneously.
Different instruments are recorded on each track, then run in parallel to
create the equivalent recording of a conventional multitrack tape recorder.
Engineers can then combine the separate channels of sound into stereo, just
as they would from multitrack tape.

To record sound with the quality associated with compact discs, the
original analogue waveform of sound is sampled at 44.1 kilohertz. Each sample
is then described by a 16-bit digital ‘word’ that generates a stream of
data of 700 kilobits per second per channel. A megabyte of storage, either
on RAM or Winchester hard disc, is needed for every 11 seconds of mono sound,
or 5.5 seconds of stereo. If the system is recording 16 tracks, each megabyte
can hold only around half a second of sound.

Professional recording is often carried out at a sampling rate of 48
or 50 kilohertz, which increases the number of bits of information that
must be stored to 800 kilobits every second, and reduces recording time
per megabyte to 10 seconds per track of mono sound.

Although chip manufacturers are promising to bring out chips that can
store 16 million bits of information, the technology is still at an experimental
stage. NED uses banks of 256-kilobit chips as the cheapest option, so its
systems need 32 individual chips to store 1 megabyte of information.