9. Presentation and delivery How you want your audio presented to the user and how you want to incorporate it into the digital resource as a whole will determine the ultimate form of your digitised sounds. There are several ways in which you could deliver a digital resource, and within those there are various ways in which the audio itself can be delivered. Your resource may be on CD-ROM, an organisation's internal network (or intranet), or the web (aka the internet). These are discussed below. 9.1. Cd-rom The benefits of delivering a digital resource via a CD-ROM are: access can be controlled easily. This may mean that copyright clearance is easier to negotiate, as the number of users can be carefully restricted. audio can be incorporated without needing to consider the operation of the resource over a network. Download time is not a factor, so the performance of your resource can be more easily assured: the delivery is independent of network conditions. This may also mean that full-bandwidth uncompressed sound can be used (for instance in the form of 'wav' files), achieving maximum sound quality. The disadvantages are: CDs are a physical medium: they have production costs, they take up space, and they need to be physically delivered to users. they have limited storage capacity, so the amount of audio you can present is limited also. they cannot be updated once created. This limits the life-span of the CD if the resource is likely to be subject to ongoing development or updating. Materials presented on CD-ROM can be organised and constructed in the same way as websites, or they can be presented as 'multimedia applications', effectively programmes which run when the CD is inserted into the computer. There are many tools (commercial and non-commercial) which assist with building both types of presentation. Some require a significant level of technical knowledge (in such areas as HTML programming, for instance), whereas others operate on a more intuitive wysiwyg (what you see is what you get) basis. 9.2. Networks and the web The advantages of presenting a resource via a network or the web are: The resource is highly accessible internationally (at least on the web). It is instantly updateable by the resource owner. There is effectively unlimited data capacity, though storage space obviously costs. It can link to other databases/resources. No physical media are involved in delivering the resource. The disadvantages are: Resource performance is limited by network performance: sound files may need to be compressed and quality may be compromised. The resource requires ongoing technical support. Having open access will impinge on copyright negotiations; alternatively, a system of access control can be incorporated. 9.2.1. Delivering audio via networks and the web There are two options for delivering audio files over networks (including the internet/web): File download: the soundfile resides on a server computer belonging to the resource owner (quite possibly the same computer that hosts the website itself). When a user clicks on a link to the sound in their web-browser, the whole file is sent from the server to the client (the user's computer). The user can usually choose to save the file to disk to play it later, or have their browser load it straight into a sound-player programme: in either case, the user will only be able to hear any sound at all once the entire file is downloaded. Streaming: again, the file resides on the resource owner's server. When the user clicks a link to the sound, this starts up a browser plug-in which downloads the file bit by bit, playing each bit as it goes (this is called buffering). The user starts to hear sound as soon as the first little bit of the file is received: usually after a second or two, no matter how long the entire file is. The plug-in will just keep receiving and playing until the end of the sound is reached. This method means the user does not have the option to save the whole file (i.e. to make a copy of it on their own computer's hard-disk). 9.2.2. Data compression With both these methods of audio data delivery, efficiency is dependant on the available network 'bandwidth'. If the network is fast, data can be sent more quickly than if it is slow. The speed of a network, including the internet, varies with the amount of other use it is getting; UK access to the internet, for example, slows down considerably when the USA 'wakes up', as there is much more network traffic. In order to minimise the problem of slow delivery, data compression is used. There are various ways in which the size of any kind of digital data can be reduced. With audio data compression, there are types of compression which are 'lossy' and others which are 'lossless'. Lossy compression discards some data and retains that which is thought to be essential. Some very effective compression methods use lossy compression to achieve huge reductions in data size but with very good estimations of what our ears need to hear and what they don't. They use a detailed mapping of the responses of our hearing system as the basis for deciding what is essential data and what is redundant. For the 'file download' method of delivery, using a compressed soundfile format can greatly reduce the time it takes to download the sound. Using modern data compression such as MPEG Layer III encoding (commonly known as 'mpeg 3' or 'mp3'), the size of a full-bandwidth soundfile can be reduced by as large a ratio as 10:1 or more, with little or no perceptible loss of quality. This results in a comparable reduction in download time. If some loss of quality is acceptable, and the file is converted to a lower bandwidth and mono, ratios of 200:1 can be achieved. 9.2.3. What difference does compression make? To demonstrate this, we need to do some maths. Standard WAV and AIFF format sound files (among others) do not generally use data compression. Every sound sample is stored as a numeric value - there is a sample value for every sample period, i.e. at a sampling rate of 44.1kHz 1 second of sound will generate 44100 sample values. If these samples are stereo and are at 16-bit resolution, then each one will occupy 4 bytes of computer memory. So our 1 second sound will occupy 44100 x 4 = 176400 bytes. A minute of sound will occupy 10 584 000 bytes, or ca. 10 megabytes of memory or disk space. Can we estimate how long it might take to download a file of this size via a modem over the internet? A modern modem might achieve a transfer rate of ca. 30kbps (kilobits per second). There are 8 bits per byte, so 1 minute of sound is 10 million x 8 = 80 million bits. 80 000 000 / 30 000 = 2666 seconds, or ca. 45 minutes. But if we use a 10:1 compression ratio, this reduces to 4.5 minutes. Hence the attraction of data compression! 9.2.4. Streaming Audio streaming depends on the server and network being able to supply enough data fast enough to the user's computer to enable it to play back the sound without audible gaps. The ability of the network to accomplish this is variable, especially where the internet is concerned; at times of heavy traffic, congestion means that much less data can get through to any particular computer in a given time period. In order to get around this problem, audio streaming software uses a kind of data compression where the compression ratio can be varied. When network conditions are good, a lower ratio can be used, giving greater audio bandwidth and better quality; when conditions are bad, more compression is used, and audio quality is reduced. The aim of this is to keep the audio coming, even if it's of poor quality. In order to stream audio, a computer must be set up by the resource owner to act as a server: its job will be to serve out audio on demand across the network. This needs to be a high-performance machine, and will require special audio streaming software. One of the commonest proprietary types of audio streaming on the web is currently RealAudio, from Real Networks Inc. For the user to hear RealAudio, they need a RealAudio plug-in for their browser. This plug-in communicates with the server software and downloads the sound bit by bit, playing it back as it goes. To set up a streaming server will be fairly costly and will require specialist computing skills. 9.2.5. Usage considerations The method of audio delivery you use will depend on what you think users will need to do with the sound. If the purpose of the digital resource is to provide a source of optimum quality materials from which further copies are to be made, then the audio should be as close to the original as possible. It should probably not use compression, because even if the losses are seemingly imperceptible, some data is still lost; further processing at a later date may yield audibly different results than if the digitisation had been done without data compression. In this example, users will probably want to store the sound locally on their hard disk, in order to be able to make copies: so the sound should be downloadable. If further copying or production work is not the aim, but the sounds are just there for people to listen to, then data compression (such as 'mpeg3') is appropriate: as discussed, this can greatly reduce download time with little or no perceptible loss of quality. If you want users to be able to click an image or a link and hear sound immediately, they don't need a copy of the sound on their hard disk for later use, and sound quality is not of major importance, then streaming is most suitable. However, this is a resource-intensive method of delivering audio. Top of Page
| 
Page last modified: by