1. Introduction
The wide-ranging scope of this book illustrates the great variety of disciplines and institutions which owe much to sound recording or archives for their essential stimulus. Indeed, without sound recording and sound archives, it would be hard to imagine some disciplines ever existing in their modern form. Although the problems peculiar to each individual subject are and will remain the prime concern of scholars in those fields, all need to consider the basic technical requirements and the physical conditions necessary if acoustic source material is to be produced and to remain unspoilt for posterity.
This chapter, therefore, is aimed at the technical layman - such as the scholar drawing up a research programme or planning a sound archive - so that he might have some idea of the technical and financial implications of his plans and to ensure that, from the very beginning, he seeks the co-operation or at least the constant advice of technical experts. 1
If we are to establish what problems have to be overcome in sound recording and sound archives, we must first know something about the nature of sound itself. The term 'sound' is generally understood to mean periodic oscillations in air pressure, but for our purposes (with the exception of some bio-acoustic matters) it is sufficient to consider sound simply in terms of human hearing. In humans, hearing sensitivity starts at 16 cycles per second (Hertz, Hz) and extends to an upper limit of 20,000 Hz (20 kHz). Each doubling of audio frequencies is called an octave and the range of human hearing covers about ten octaves. With regard to different sound levels, human hearing has an extensive range. Hearing sensitivity begins at a sound pressure level of 2.10-4 micro-bars and extends to as high as 6.102 micro-bars, at which point the sensation of hearing becomes painful. The ratio of these sound pressure levels, which we term the 'dynamic range', is about 1:3,000.000 or l30dB2. All acoustic phenomena, whether a tone, a sound, exotic music, speech or any sort of noise, are made up of quite individual patterns of oscillation each of which can comprise a variety of partial oscillations but which can all be expressed in the fundamental parameters of frequency (= cycles per second or Hz) and amplitude (sound pressure level, expressed as dB).
By a transformation process, sound recording has to convert these parameters into a medium in which the function of time is rendered as a function of space, i.e. every moment in a sonic event has to correspond to a particular point on a recording.
Old phonographs and the early gramophones used to do this mechanically. Vibrations in the air were captured in a horn, exciting a membrane which in turn set in motion a stylus which left modulations on a revolving cylinder or plate, each modulation corresponding to the vibrations of the membrane and the air. The process of reproducing the sound simply involved reversing the procedure. The modulated groove moved a needle which then caused a membrane to vibrate and the vibrations would be amplified and made audible by the horn. In time electronic methods of recording and reproduction came to be used; the air vibrates the membrane of a microphone which then transforms the vibrations into an analogous alternating current. After precise amplification, this drives an electric cutting head. In reproducing the sound, we nowadays use electro-dynamic or electro-magnetic sound pick-ups which convert the modulation of the record groove into an analogous alternating current. By means of an amplifier, this will then drive a loudspeaker or headphones, which convert the alternating current back into air vibrations. The disc itself has not changed in principle since it was originally introduced.
In magnetic sound recording, the sound is stored on a magnetic tape. The alternating current from the microphone is converted in the machine's recording head into an alternating magnetic field. The tape, moving at a constant speed across this recording head, has each particle stored within its coating re-arranged in accordance with the alternating field. To reproduce the sound, the magnetic field fixed on the tape produces an alternating current as it passes across the replay head and, suitably amplified, is converted into sound by the loudspeaker. The process functions that much more precisely the more linear the system or, to put it another way, the less distortions there are. These distortions can be classified as three different types:
- Linear distortion: this is caused by uneven sensitivity at different frequencies. The ideal requirement is a flat frequency response i.e. constant sensitivity across the whole audible range, from about 20Hz to about 20,000Hz.
- Non-linear distortion: this is caused by changes in the original wave form and appears as spurious sound not present in the original (harmonic distortion, intermodulation distortion).
- Modulation distortion: distortion mainly caused by irregularities in the tape travel. Pitch fluctuations ('wow' and 'flutter') and 'mod-noise' come under this heading.
In addition to these distortions there is the unavoidable noise which we find in every link in the chain and which occurs particularly in the sound carrying medium itself, limiting the dynamic range which can be stored. Despite all the technical progress which has been made since sound recording was invented, there is still no system so perfect that distortion will not occur. Even though there are no difficulties nowadays in converting and storing the whole audible frequency range, non-linear and modulation distortions are still with us. More important, with analogue recording techniques only 60 of the 130dB dynamic range of human hearing can be stored (although with the introduction of digital technology the upper limit of this can be raised to 96dB).
In view of what we shall see can be quite considerable outlays for machinery, servicing, tapes and their care and storage, we must explain why this cost-intensive standard is necessary for sound recording and archiving. In contrast to the written and printed word which reproduces a verbalised mental process by a series of representational symbols, a sound recording documents a physical event which can be repeated at any time after the event itself. A certain amount of redundancy is intrinsic in speech and writing and - without any real detriment to communication - letters, words, even whole clauses can be omitted. The essential value of a sound document, however, lies in the very information which it supplies over and above what can be transcribed; such as form and variations in tone, manner of speech, or - in the field of music - the timbre, the performance, the subtleties of rhythm. Here, too, we see more clearly that musical notation provides no more than a framework, just a small part of the total musical message. In the case of a noise, however, written symbols cannot provide an adequate substitute for any part of a sound recording. It is the very information which a transcription simply cannot convey, which provides the criteria and the fundamental argument for high quality sound recording. If, for any reason, the additional information supplied by a sound recording is considered worthless, then the recording is no more than an intermediary substitute for a written record and is not worth keeping. Merely making a voice intelligible or a melody clear enough to be transcribed should not become the sole criterion for technical standards in sound recording. Rather, it should ideally be a question of exhausting all available means to obtain the optimum quality of recording. The value of this ideal is further underlined by the fact that more and more scholars in various fields (musicology, linguistics, psychology) are endeavouring to make this element in sound recording which transcends writing, a subject for serious research.
It must not be concluded, however, that an ideal recording technique will necessarily produce an ideal recording in practice, for an array of technical gadgetry may end up distracting and disturbing the subject being recorded. Careful consideration must be given in every case to producing a compromise between the requirements of technical precision and the need for equipment to be unobtrusive, together with all the implications this may have for the choice of equipment and the resulting recording standards.
The question as to whether existing historical recordings are worth retaining despite their technical shortcomings is something which must be decided after consideration of the contents in each case, but a decision to retain them should not lead us to the conclusion that we may abandon the idea of trying to obtain the highest possible standard of recording in new projects.
- The Technical Committee of IASA is preparing a Technical Manual for Sound Archives which will deal in depth with all the technical problems specifically related to archives and will be aimed at a technical readership also. As this chapter is only meant to serve as a basic guide, it does not go into detail and deals only with conventional analogue recording techniques. For more detailed information, readers should consult the literature listed in the bibliography to this chapter.
- dB (decibel) is logarithmic ratio measurement of sound pressures or its analogous voltages.