Why can the difference in bitrate make it sound great (high, medium, low)?
/Bit-Depth-vs-Bit-Rate-d62ed553edc644aab33a41d78f2c4419.jpg)
Reply:
Just to make sure this is clear, let’s differentiate
sample rate vs bit depth
as much as
Bit rate
how they relate to audio in the digital domain …
Sampling frequency:
The sample rate is specified as a frequency (samples per second), for example, 44.1 kHz for CD. Other common values are 48, 88.2, 96, 176.4, and 196 kHz, although some formats (such as DSD) have sample rates greater than 2.8 MHz. The sample rate indicates
how often the audio signal is measured
While some people view lower readings as a tiered bar graph, I prefer to view them as a child bitmap. If you take the outline of a horse and simplify it to 20 points so the child can connect, it’s not so much that you end up with steps (using straight and curved lines to connect 20 correctly spaced points can lead to a decent figure), but there won’t be without subtlety. Whereas with 200 (or 2000) points, you could approximate the wavy strands along the horse’s mane.
In audio, a lower sample rate does not make the sound “bad” (eg, fuzzy, fuzzy, or distorted), but rather limits the maximum frequency (pitch) that can be recorded / played back as intended.
Nyquist theorem formula
, The 44.1 kHz sampling rate was chosen for CD because it can record and play back frequencies up to 20 kHz. To record a spoken word (such as a speech, a sermon, or an audiobook), it would be difficult to detect a much lower sample rate, as the human voice has less and less harmonic information above 10 kHz.
Depth bits:
Considering that the sampling frequency determines how
often
audio signal is measured, bit depth indicates
scale accuracy
Since we are talking about digital audio, we describe this measurement scale in bits, where each bit is 0 or 1, and we concatenate a certain number of them to represent the value. When we have 8 bits, there are 256 possible numerical values, including zero. With 16 bits, there are 65,536 possible values. A 24-bit register can use 16,777,216 values.
When we convert analog audio to digital representation (A-to-D) and vice versa (D-to-A), we find interesting mathematical relationships. Each bit (digital) doubles the number of possible values … And doubling the amplitude (approximately 4 times the power) of the sound wave (analog) corresponds to + 6 dB of loudness. Therefore, we can estimate the maximum dynamic range * of a digital recording at 6 dB / bit. Therefore, 8-bit recording has ~ 48 dB of dynamic range, 16-bit recording (such as a CD) has ~ 96 dB, and 24-bit recording has ~ 144 dB.
* For those of you unfamiliar with this term, dynamic range basically describes the difference between the quietest and loudest sound waves that can be recorded / played back. The CD has a difference of approximately 96 dB, which can be used to represent the most subtle pause compared to the incredibly loud burst of the cannon at Tchaikovsky’s climax.
1812 Overture
,
Three quick notes for those interested in delving into the rhythm …
There is a formula for the actual dynamic range of a digital recording that may differ slightly from the previous estimate, but it is a fairly minimal deviation, so an estimate of 6 dB / bit is what you normally see in quotes.
The latest 32-bit floating point representations combine a 24-bit number and an 8-bit exponent to represent many more possible values than 24-bit registers. The dynamic range estimate is getting a bit dubious, but suffice it to say it’s well above 144 dB.
Using a lower bit depth, while you might think in terms of warp plugins with names like “bit-grinder”, doesn’t have to sound “bad” (eg fuzzy, fuzzy, or distorted), but just represents a reduced dynamic range. But since a 16-bit recording with a dynamic range of 96 dB (65,536 numerical values) cannot be represented in 8 bits (48 dB and 256 numerical values), to reduce the bit depth of the already digitized audio, a mathematical correction of the numbers down. (for example, 65535 becomes 255) using a compressor or limiter, which can cause the quietest recording bits to be lost so that the difference between soft and loud parts is <48 dB. Without such scheme, the transformation will cause clipping (numerical values above the maximum),
Bit rate:
In digital audio, the bit rate is a measure of
how many bits are transmitted / processed per second
