Analog sound vs. digital sound

From Wikipedia, the free encyclopedia

This article compares the two ways in which sound waves are recorded and stored. A sound wave can be thought of as a signal which over time can vary continuously in amplitude. This signal can be recorded either digitally or in an analog format. An analog recording is one where the original sound signal is modulated onto another physical signal carried on some media or substrate such as the groove of a gramophone disc or the iron filings of a magnetic tape. A physical quantity in the medium (e.g., the intensity of the magnetic field) is directly related, or analogous, to the physical properties of the sound (e.g, the amplitude, phase, etc.).

A digital recording does not attempt to record all of the original, continuous signal. The sound wave is first converted into an electronic signal by a microphone. Measurements of the signal intensity are then made at regular intervals (sampling). At each sampling point, the signal must be assigned a specific intensity from a set range of values (quantization). For example: the signal being measured is only allowed values between 0-255. If the signal has an intensity between 233 and 234, e.g. 233.5 it must be assigned a value of either 233 or 234. Using a greater range of possible assignments, e.g. 5000 instead of 256, increases the accuracy of the measurement. Sampling (i.e. measuring the signal) with greater frequency also increases the accuracy. In doing this, the original sound wave can now be described using only numbers. When playing a digital recording, the numbers are converted back into a continuous, analog signal. This electronic signal is then amplified and converted back into a sound wave by a loudspeaker.

[edit] Analogue sound

For a full treatment of the history of sound recording, see the article Sound recording and reproduction.

The earliest sound recordings and broadcasts were analogue. The technology to record sound was invented in the late nineteeth century. Radio became popular in the US in the 1930's. The sound quality of these broadcasts and recordings was sufficient to reproduce speech but were poor for music. Treble and bass were not reproduced well, and music having a large dynamic range (the difference between the softest and loudest parts of a performance) like classical symphonies, lost their dramatic impact. This is because the recording equipment could not accurately capture the relative intensities of the different sections of the music.

With the invention of the long playing record (LP for short), which became popular in the late 1950's, sound quality for the consumer was beginning to approach what would now be considered high-fidelity. The twelve inch disc was popular all the way into the 1980's. Latterly it had a playing time of up to thirty minutes per side, making a single disc long enough for most pop music albums and classical symphonic works. The other popular size was the seven inch, with a shorter running time ideally suited for pop music single releases.

Despite the continuous improvement to both sound (and video) analogue formats, they still had a number of drawbacks:

• Analogue systems suffer from a noise floor that varies between different machines.
• With the exception of vinyl records and Laserdiscs, most analog media are tape based with sequential access.
• All analogue audio systems are subject to electrical and mechanical noise when reading, recording and processing sound.
• Analogue systems can be subject to wow and flutter, unless they are somehow timebase-corrected like VHS tapes.
• High-quality open-reel tape and related hardware is expensive to buy and maintain.
• Each analogue-analogue copy or regeneration yields inferior quality, especially for standard consumer-grade equipment.

[edit] Vinyl 'warmth'

Compared to CD, vinyl records are sometimes said to sound 'warmer', and this is something that appeals to some listeners. This warmth can be explained in the limited high-frequency response of the vinyl format, which deteriorates with each playing of the record. This is due to the wear of the stylus in contact with the record surface. When a CD is played, there is no physical contact involved, and the data is read using a laser beam. Therefore no such deterioration takes place, and the CD will, with proper care, sound the same every time it is played.

Sony engineers have stated that the lack of warmth of the audio CD format is related to the use of analogue-to-digital and digital-to-analog converters, and the sharp filtering of frequencies above the audible range (Maes & Vercammen 2001). The Direct Stream Digital format, developed by Sony, is meant to sound better because the filtering and conversion used in the PCM format (the way signal information is stored in audio CD) is not required. This claim of improved sound quality has not been demonstrated in properly controlled subjective tests and is not generally accepted.

[edit] Digital sound

Up until fairly recently, sound was recorded and replayed only on analog equipment. Technology was not sufficiently developed to allow for high quality digital recordings. For the most part, recordings were made on tape and then distributed on vinyl records. The earliest demonstration of high fidelity digital audio technology was in the 1960's (Maes & Vercammen 2001), with a recorder that used the digital PCM format. It was not until the late 1970's that digital recorders were widely used by record labels.

Digital audio recording was developed because it was realised that analog technology had practically reached its highest limit in sound quality. Analog equipment suffers from noise and distortion which cannot be entirely removed. Additional noise and distortion of the original signal is added every time a recording is copied. Through careful design these aberrations can be reduced to low levels, but they will always be present. Digital audio is more resistant to noise and distortion. This is because when the original signal is converted into binary numbers (1's and 0's, called 'bits') further additions of noise and distortion (in the form of digital errors) can be rejected at every stage of processing. The information is still contained on an analog, continuously variable medium such as magnetic tape but since the signal is known only ever to be 'on' or 'off', it is easier to reject any noise (errors) that are added.

A digital recording is produced by first converting the physical properties of the original sound into digital information which can then be decoded for reproduction. The conversion process can be susceptible to noise and imperfection. However, the nature of the physical medium is immaterial in recovery of the encoded information as long as the individual bits can be recovered. A damaged or dirty digital medium, such as a scratched compact disc may also yield degraded reproduction of the original sound, due to the loss of some digital information in the damaged area. Error correction codes have been developed to combat data loss due to physical imperfections of digital media.

[edit] Why digital?

Recording and storing sound or video digitally has a number of advantages over analogue methods, including:

• Most digital media have non-sequential (random) access, due to their disk or memory-based nature. However, there are sequential access digital media such as DAT and DCC.
• Consumer standard digital sound (16 bit) yields a signal-to-noise ratio (SNR) of 96 dB, which is generally superior than even the lowest analogue noise floors. The SNR is even lower as the bit depth is increased.
• Copies and regenerations can be exact clones and be made infinitely without degradation, unless DRM restrictions or mastering errors apply.
• Generally higher resistance to media deterioration.
• Ability to apply error-correcting codes to prevent data loss and/or corruption.
• Ability to encode non-audio information into the audio stream such as information about the owner, track titles, etc.
• Easier multiplexing of multiple channels into one data stream, something impossible on most analogue media.
• Ability for the same medium to be used with arbitrarily high or low quality encoding methods and number of channels or other content, unlike mechanically pre-fixed speed and channels of practically all analog systems.
• High quality sound output is achievable with very low manufacturing cost and in mass-produced devices.

The use of digital audio equipment can be an entirely transparent process. This can be shown using subjective test methods [1]. High quality digital recorders have excellent signal-to-noise ratios and a wide frequency response. Their performance can be reliably measured using objective tests [2]. Digital tape recorders, with their improved performance characteristics, were adopted early by most of the major recording labels. The digital recorders developed by record companies and broadcasters often had a level of performance far exceeding that of analogue recorders (Driscoll 1980). The ability to record, edit and mix recordings digitally did away with creatively restrictive analogue production techniques like direct metal mastering.

Digital audio can be thought of like a 'window' that encompasses the original signal. As long as the window is big enough to record the peaks and lowest details of the original signal, then the digital process is entirely transparent. The fact that the signal has been through this process cannot be shown by any test, unless that test is somehow flawed (Rumsey & Watkinson 1995).

[edit] Criticisms of digital

• Low bit depth for signal level quantization, typically anything below 14 bits, can lead to significantly harsher sound output, considering 80 dB of SNR as an informal "minimum" for Hi-Fi audio. However, it is uncommon to find digital media specified for less than 14 bits, except for older 12-bit PCM Camcorder audio. The industry standard for CD quality audio is 16 bits.

• If digital audio data compression is used excessively, then reduction in sound quality will occur [3]. The audio CD format does not use compression, and so is not affected by this problem.

The following disadvantages should not be a problem with correctly operated, properly designed equipment:

• Maximum audio frequency response is "hardcoded" by the sampling frequency itself. A digitized audio waveform can exhibit audio frequencies up to half the sampling frequency, which is also called the Nyquist frequency. Any frequency content above the Nyquist frequency is filtered out in most cases, or otherwise aliased. This "hardcoding" is carried over to output filters and oversampling units as well. The industry standard for CD quality audio is 44.1 kHz, which has a Nyquist frequency of 22.05 kHz. Analogue audio does not require the use of these filters, potentially allowing for a wider frequency response

Digital audio can have a sufficient frequency response to cover the entire audible range. The design of filters used to cut off frequencies above the limit of the particular digital format (~22 kHz for audio CD) should prevent aliasing being audible. While analogue audio is unrestricted in its possible freqency response, the limitations of the particular analogue format will provide a cap. Vinyl records may record signal information above 20 kHz, but fidelity will be lost on each playing. CD's do not suffer this fidelity loss, and have a frequency response covering the whole audible bandwidth. At the lower end of the frequency range, vinyl records suffer from rumble, restricting low bass below ~30 Hz.

• In digital recording, quantization of the original analogue signal results in quantization noise. Unlike the analogue noise floor, quantization noise is non-random in nature, and is more audibly disturbing.

In digital audio, dithering is used to hide subjectively undesirable quantization noise. The effect of using dither lowers the signal-to-noise ratio of 16 bit from 98 dB to around 90 dB (Maes & Vercammen 2001). Use of dithering hides quantization noise and adds a small amount of white (random) noise to the original signal. The amount of noise will vary depending on the format, but the signal-to-noise ratio of digital equipment is usually better than for equivalently priced analogue equipment. The vinyl format usually is limited to a 60-70 dB noise floor, which is the analogue equivalent of signal-to-noise ratio.

• Presence of a "hard" recording lower level, which may lead to audible clicks and static instead of a white noise low level hiss.

As long as the digital recording equipment is set up correctly, this should not happen. 16 bit digital has a signal-to-noise ratio covering 90 dB, which should be enough to preserve all low level musical detail along with the peak music transients. Dithering can increase the dynamic range of 16 bit recordings above 90 dB.

[edit] Quantization and very low signal levels

Some sustainers of analog sound claim that there is no hard "floor" (lowest sound level) beneath which recording is not possible. Instead, the desired signal simply slips farther and farther into the noise floor as its amplitude is reduced.

This statement is suspicious and could only be true for analog signals that are strong enough to be above the (unavoidable) mechanical, electrical and thermal noise level in the recording and playback cycle (mechanical transducers (microphones, loudspeakers), amplifiers, recording equipement, mastering process, reproduction equipment, etc) .

It makes in fact little sense claiming that an analog signal can "use" all the available physical resolution of a medium and be accurately recorded, when that same signal can be weaker, at low levels, than the sum of all external noise, interference, and unwanted signals that are recorded at the same time. This applies of course to both analog and digital systems.

Mathematically, this can be expressed by means of the signal to noise ratio. On an 8-bit digital system, there are only 2^8= 256 possible signal amplitudes, of which there are 256 discrete amplitudes relative to the minimum signal level, which results in a dynamic range of just 48.165 dB, which is inferior to most cassette tape systems, so in fact 8-bit recordings tend to sound noisy and scratchy, and miss low-level signals.

• Note that a decibel is one-tenth of a Bel. It is a somewhat strange concept that characterizes the logarithmic nature of human senses. Now to make it more complex, the amplitudes discussed in this article are voltage levels. To convert a voltage level ratio to a Bel, simply divide them and calculate the logarithm to base 10. Then multiply by 10 to get decibels. Unfortunately, Ohm's Law comes into play; the power of the sound is approximately the square of the voltage level. Summary: The human hearing range is around 120 dB. A digital recording has, at best, a range of 20 * log10 (2 ^ number of bits).

On the other hand, a system with a 16-bit quantization has a dynamic range of 96.33 dB, which is generally considered Hi-Fi and way beyond the signal to noise ratio of most consumer audio systems, and it's difficult, in practice, to find an analog sound recording system that can offer a better sensitivity at a reasonable price and implementation complexity.

In practice, each additional quantization bit adds a notable 6 dB in signal to noise ratio, e.g. 144 dB for 24 bit quantization (24 x 6 = 144), which is however very rarely (if ever) achieved in practice, with 21-bit (126 dB) and 20-bit (120.4 dB) being more practical, see DAC and ADC for more details.

To make a comparison, cassette tapes are generally below 70 or even 60 dB; FM broadcasts are more or less the same; an average vinyl record, if in good condition, can sometimes surpass 85 or 90 dB and a properly mastered CD can approach or even exceed 90 dB.

For example, a 0.5 V peak to peak input line signal, quantized at 16-bits, would require an equivalent minimum input sensitivity of just 7.629 microvolts, or an equivalent 15.3 ppm sensitivity by part of the whole recording system and medium, which is only achievable with studio-grade equipment, perfectly crafted and preserved medium, and cannot be achieved during reproduction by the majority of consumer audio systems, at a physical-electrical level.

[edit] Early digital recordings

Many of the criticisms levied against digital sound reproduction stem from the early days of the technology. Due to the extra resolution and sound quality afforded by the audio CD, faults in the recordings were easier to spot (Greenfield 1986). Background noise, inadequate acoustics and poor microphone placement could be heard far more easily than on vinyl records. Some recording engineers like Jack Renner of the Telarc record label were more aware of these problems than others, and early on were able to produce recordings of excellent sound quality. Analog sound reproduction was already a mature technology when digital recording and compact discs first appeared. It should be said however that even first-generation digital recorders had performance characteristics matching those of the best analog recorders (Driscoll 1980). Audio professionals also needed some time to build a body of knowledge, as analog techniques could not always be directly transposed to the new digital medium.

Much progress has been made since. Progress in electronics and economies of scale from mass production of CD players led to improvements in Digital to analog converter technology. Professional digital recording equipment correspondingly improved to even higher levels, and digital equipment having accuracies up to 18 bits (and higher) are not unusual.

[edit] Shape of the waveforms

Proponents of analog recordings argue that it is superior to digital for the reason that digital recordings are an approximation of a waveform. Analog recordings could also be considered as approximations of the original signal, since they will inevitably add noise and distortion to it.

In digital systems, the waveform only needs to be sampled at a sufficient bit depth and sampling rate so that the added noise and distortion is not audibly intrusive. In 16-bit audio, there are total of 65,536 different values can be assigned or quantized to each sample. This format allows for a signal to be reproduced with very low levels of noise and distortion. Sampling the waveform at higher frequencies and using a greater bit depth allows noise and distortion to be reduced even further. DAT can store audio at up to 48 kHz, while DVD-Audio can be 96 or 192 kHz and up to 24 bits resolution (as an aside, in the movie industry, the choice to use a 48 kHz sampling rate is at least partly due to the easier time relationship with the 24 frames per second film format).

The improvements in signal quality from doing this must be weighed against the fact that human hearing is not perfect. For example, listening tests have shown that levels of harmonic distortion in an ordinary listening environment are tolerated in whole numbers, although it may fall to lower levels (0.05%) depending on programme material (Toole et al. 1994). What must also be considered is that other parts in the reproduction chain will usually add more distortion and noise than the digital audio device. Loudspeakers, for example, in common with other transducers, typically have a more uneven (and limited) frequency response than wholly electronic devices like CD players.

Digital audio equipment is designed so that quantization noise and aliasing are not problems. As has been stated earlier, quantization noise is hidden using dither, which slightly worsens the signal-to-noise ratio but improves the subjective sound quality. Whether or not aliasing occurs depends on the quality of the filter. Understanding the use of filters in digital audio requires some background. The choice of sampling rate used in a digital system is based on the Nyquist-Shannon sampling theorem. This states that a sampled signal can be reproduced exactly as long as it is sampled at a frequency greater than twice the bandwidth of the signal. Theoretically then, a sampling rate of 40 kHz would be enough to accurately reproduce a signal having frequency bandwidth up to 20 kHz. The difficulty arises in removing all the signal content above 20 kHz, and unless this is done, aliasing of these higher frequencies may occur. This is where these higher, inaudible frequencies alias to frequencies which are in the audible range. To prevent aliasing, it is not necessary to design a brick-wall filter, that is a filter that perfectly removes all frequency content above (or below) a certain range. Instead, a sampling rate is chosen above the theoretical requirement. This allows for a less severe filter to be used. In addition to this, other methods can be used to increase performance, for example, oversampling.

[edit] Scientific and technical validity

The matter of whether analog sound can truly be superior to digital sound is usually considered pseudoscience and looked upon by people with a solid technical education, including but not limited to engineers, physicists, etc.

This is not without a reason, as in practice digital signal processing systems have been amply proven to reliably transmit, handle, store and process many different kind of signals, ranging from control signals and biomedical data, to video and images, also including audio, while their reliabilty, performance, unit cost and ability to be mass-produced are also in general much superior to those of equivalent analog systems, so it would seem absurd to imply/assume that audio cannot be handled as well as e.g. video.

Regarding audio, a suitably-specified digital audio system (for a given frequency bandwidth and a given SNR should have, in theory, no audible differences from an equally specified analog system, provided that reproduction is performed correctly.

On the other hand, many audiophiles suggest that an analog sound system somehow manages to convey more information than what most digital media can (usually in the form of an ultrasonic frequency response, and with different kinds of distortion that lead to a supposedly more pleasureable listening experience. However, assuming off-specification frequency response violates the clause that two comparable systems should be "equally specified".

[edit] Was it ever entirely analog or digital?

Complicating the discussion is that recording professionals often mix and match analog and digital techniques in the process of producing a recording. Analog signals can be subjected to digital signal processing or effects, and inversely digital signals are converted back to analog in equipment that can include analog steps such as vacuum tube amplification.

For modern recordings, the controversy between analog recording and digital recording is becoming moot. No matter what format the user uses, the recording probably was digital at several stages in its life. In case of video recordings it is moot for one other reason; whether the format is analog or digital, digital signal processing is likely to have been used in some stages of its life, such as digital timebase correction on playback.

[edit] Hybrid systems

While the words analog audio usually imply that the sound is described using a continuous time, continuous amplitudes approach in both the media and the reproduction/recording systems, and the words digital audio imply a discrete time, discrete amplitudes approach, there are methods of encoding audio that fall somewhere between the two, e.g. continuous time, discrete levels and discrete time,continuous levels.

While not as common as "pure analog" or "pure digital" methods, these situations do occur in practice. E.g. while vinyl records and common compact cassetes are analog media and use quasi-linear mechanical encoding methods (e.g. spiral groove depth, tape magnetic field strength) without noticeable quantization or aliasing, there are analog non-linear systems that exhibit effects similar to those encountered on digital ones, such as aliasing and "hard" dynamic floors (e.g. frequency modulated audio on VHS tapes, PWM encoded signals).

Although those "hybrid" techniques are usually more common in telecommunications systems than in consumer audio, their existence alone blurs the distinctive line between certain digital and analog systems, at least for what regards some of their alleged advantages or disadvantages.

[edit] See also

• Audiophile

[edit] References

• Arthur, C. "Is digital radio really any better than analogue?", The Guardian, January 17, 2006
• Driscoll, R. (1980). Practical Hi-Fi Sound, Hamlyn.
• Greenfield, E. et al (1986). The Penguin Guide to Compact Discs, Cassettes and LP's, Penguin.
• Krueger, A. The PC AV Tech, 11 June, 2005.
• Lipshitz, S. "The Digital Challenge: A Report", The BAS Speaker, August-September 1984
• Maes, J. & Vercammen, M. (2001). Digital Audio Technology 4th edn, Focal Press.
• Pohlmann, K. (2005). Principles of Digital Audio 5th edn, McGraw-Hill Comp.
• Rumsey, F. & Watkinson, J. (1995). The Digital Interface Handbook 2nd edn, Focal Press.
• Toole, F. et al (1994). Loudspeaker and Headphone Handbook 2nd edn. Edited by John Borwick. Focal Press.