# COLUMN

## Glossary

### Technical term of VELVET SOUND

What does the word mean used in audio products?

Audio engineers explain basic words that help you understand audio products more and more, such as S/N and THD.

* Please note that some expressions are not always accurate from the engineer's perspective.

### dB: decibel

dB is a unit that expresses the ratio of loudness and signal strength. d (decimal) is a prefix with the meaning of 1/10. Human senses such as loudness are proportional to the logarithm of actual physical quantities. For this reason, the sound is also expressed in dB, which is the quantity to make it easier for humans to recognize.
Roughly speaking, in the audio industry, “(digits-1) x 20” is dB (see table below).

### Audible Frequency Band:Audio Measurement Bandwidth

It is said that the frequency band (audible band) that humans can recognize as sound is usually about 20Hz to 20kHz. For this reason, the audio characteristics are also specified at to 20kHz or to 22kHz.

* Incidentally, the higher the frequency (treble) becomes harder to hear as you increase the age.

### A-weighting

The frequency characteristics that model the sensitivity of the human ear are shown below. It has a peak around 2 to 3kHz. You can use this curve to compensate for characteristics that place an emphasis on how loud the noise is. S/N and DR are usually calculated with an A-weight curve.

* The low frequency part is small because it cannot be heard by humans.

### Resolution

Resolution is a measure of the fineness of digital data expressed in bits. It is easy to understand by considering it as a scale of a ruler. The 1mm scale in 30cm ruler is roughly the same as the resolution of the 8-bit data (2 to the power of 8 = 256).

Let's think about how much audio data exist. The CD is 16-bit data format (2 to the power of 16 = 65,536). If you use a 30cm ruler, the scale will be 1/10 (approx. 4.5um) of the thickness of your hair.

For 24-bit DVDs (224 = approx. 1.60 million), the size of the bacteria (approx. 18nm) is one scale. For 32-bit (2 to the power of 32 = approx. 4.3 billion) audio source would be a helium-atom-sized scale (approximately 70pm).

As you can see, audio signals are represented with levels of detail that humans cannot imagine. Just as the resolution of a digital camera increases, the image becomes clearer. The higher resolution, the higher sound reproduction and the closer it is to the “original sound.”

### S/N: Signal to Noise Ratio

S/N is calculated as “Signal / Noise” as it is read. It is calculated as the ratio of the maximum output power of a single frequency (basic 1kHz in audio) to the noise. Since how much noise is heard is an indicator, it is customary to calculate by correcting the sensitivity of the human ear with a modeled A-weight curve. Higher S/N ratios provide better audio performance, which has long been an important indicator of audio.

If this value is low, you will hear a "sar" sound (so-called "white noise") that is unrelated to the sound source. This means that it is difficult to hear a small sound source. You can listen to a "sar" sound by turning the volume to maximum while leaving nothing played on your audio device. Some systems do not hear this sound by automatic mute function. However, please be careful when testing! This is the maximum volume, so if you play the sound source, you will hear a blast!

S/N is similar to DR (dynamic range), but this is the ratio to noise when no signal input.

### THD+N: Total Harmonic Distortion + NoiseS/(N+D): Signal to Noise and Distortion Ratio

These two indicators have the same meaning and show how much of the component other than the signal (distortion component + noise component) is present. Some manufacturers only have different notations. However, when expressed as a numerical value, the sign changes in dB because the numerator and denominator of the formula are interchanged.

THD+N   = (Distortion + Noise)/Signal

S/(N+D) = Signal/(Distortion + Noise)

In any case, the larger the absolute value, the smaller the noise and distortion components.
On the other hand, the smaller this value (the larger the distortion), the more it changes from the original waveform.
* Incidentally, in the communications field, it is called “SINAD (Signal-to-noise and distortion ratio)” and is used as an indicator of communication quality.

You may be wondering why this characteristic is not corrected with an A-weight curve. In S/N and DR, noise is an important factor, and it was an index to how large it can be heard in the whole. However, THD+N is important for how much the original signal waveform can be reproduced as it is, so it does not need to be corrected with the A-weight curve regardless of the sensitivity of the human ear.

### DR: Dynamic Range

DR is used as an indicator of how small a signal can be expressed. For example, in the case of a DAC, DR is the ratio of "maximum DAC input level" to "distortion + noise" when a signal of 1/1000 (-60 dB) is input. However, the distortion is very small, and most noise is dominant. In other words, it is the ratio of "Maximum DAC input level" and "Noise at minute signal input."

It has the same meaning as S/N and is often the same as S/N. The difference from S/N is that noise of S/N is noise when there is no input, and noise of DR is noise when a very small signal is input.
In rare cases, S/N and DR are differ. S/N may be increased by changing the status from normal operation when there is no input.

* Since noise is dominant in DR measurement, that is performed with a A-Weight filter in the same way as S/N measurement.
* DR is calculated using THD+N at -60dB output. For example, suppose that the THD+N is -36dB at -60dB output. The DR is then represented as -60dB + (-36dB) = -96dB.

### Inter-channel Isolation

This is also known as Crosstalk. At low values, for example, the sound coming from the stereo right channel is heard from the left, while the sound coming from the left channel is heard from the right. This represents the amount of leakage from one channel.
Experimental results show that if this characteristic is deteriorated to 30dB or less, humans will understand the difference. A typical audio IC has 90dB or more, which is a level less than 1/1000 that humans can perceive. Normally, this is not a problem.

### Inter-channel Gain Mismatch

This shows how much the output is different for each channel in the chip. For most products, this value is very low. It is uncomfortable to listen to the difference in sound volume between the left and right of a stereo or earphone. Some amplifiers allow you to change the left and right balance, but it is not usually adjusting the left and right volume separately.