Since the introduction of the compact disc audio (CD) and the advent of digital audio tape (DAT), digital technology has become increasingly popular in the audio industry. Both CD and DAT use pulse code modulation (PCM) as a basic digitizing process. This technology translates the original analog audio signal into the digital world through sampling, quantization, and encoding. Since PCM does not use data reduction, excellent sound quality is achieved, but is purchased at the cost of high memory requirements. In PCM, a CD can contain a maximum of 80 minutes of audio data.
Why reduce the audio data?
The high memory requirements of PCM, in particular, made direct use of this technology in multimedia or digital radio systems ineffective, time-consuming or impossible. These systems require a radical thinning of the audio signals. The reasons for this are insufficient broadcast transmission capabilities, the limited transfer rate of current bus systems (PCI, IDE, SCSI) and, above all, the still lack of storage space. Not only is there a shortage of hard drive space, but the main memory in today’s PC systems also offers insufficient reserves to allow sensible work with PCM audio data. If you consider that a 6-minute piece of music in PCM requires up to 60 Mbytes of memory (WAV file), it is easy to imagine that streaming this piece, for example over the Internet, is not profitable. not to mention classic works that last several hours. The result would be extremely long download times.
On the other hand, digital technology has unbeatable advantages over analog technology. Very good sound quality, immunity to interference and relatively easy technical manageability were reasons enough for several research institutions to develop more and more methods in recent years that allow to reduce the storage requirements of digital audio signals and, therefore, its use in new areas such as digital broadcasting. The main objective was to maintain sound quality, using the CD as a reference. The result is a whole series of codecs, some of which save a considerable amount of data. At the moment, the MP3 codec, developed by Motion Pictures Expert Group (MPEG), which is widely used on the Internet, is probably the best known, but also MPEG 2, AC-3,
The amount of memory required by a digital audio signal is primarily determined by the bit rate and the sample rate. Both parameters can be adjusted while encoding the signal. The next section examines the effects of changing the sample rate and bit rate when processing signals.
According to Shannon’s sampling theorem, sampling must take place with at least twice the maximum frequency of the function to be discretized. In the audio range, where 20 kHz is the upper limit, at least 40 kHz is required. The CD uses 44.1 kHz to avoid aliasing effects. Sampling can be used to reduce the data. Lowering the sampling rate results in fewer samples that need to be stored. Needless to say, this dramatically reduces the storage requirement. Unfortunately, this tactic has one major drawback. If you reduce the sampling rate, you can easily conflict with the sampling theorem. If you wanted to sample an audio signal with the full frequency range (20Hz – 20kHz) with, for example, only 20kHz, extreme alias distortion would occur. Playing music would be completely impossible. However, sampling is sometimes a way to reduce the data rate. If, for example, only speech intelligibility is desired without high-quality music reproduction, the 20 kHz frequency range is unnecessary. 3 kHz is sufficient as the upper limit frequency. Here the audio signal can be band limited to 3 kHz with the help of a low pass filter and the sample rate can be reduced to a minimum of 6 kHz. One possible use of such low sample rates would be telephone applications, for example. Here, the audio signal can be band limited to 3 kHz
