Audio and Video Compression Basics

Free Download Mp4Gain

Audio and Video Compression Basics

As we rely more and more on digital media, understanding the basics of audio and video compression becomes increasingly important. Compression is the process of reducing the size of digital files without sacrificing too much quality. Without compression, media files would take up a lot more space on our hard drives, making it difficult to store and share them. In this article, we’ll explore the fundamentals of audio and video compression and how it works.

Understanding Audio Compression

Audio compression is the process of reducing the dynamic range of an audio signal. Dynamic range is the difference between the quietest and loudest parts of a sound recording. Compression reduces this difference, making the quieter parts louder and the louder parts quieter. This is useful for improving the overall balance of a mix, and also for preventing distortion when the loudest parts of a recording exceed the maximum level of the recording medium.

Compression can be applied during recording or in post-production, using software tools like mp4gain. When done properly, compression can improve the clarity and punch of a recording, making it sound more polished and professional. However, overuse of compression can lead to a loss of detail and a “squashed” sound that lacks dynamics.

As musician David Byrne said in his book “How Music Works”:

“A good mix is one where the listener can hear and feel everything that the musicians and the engineer intended to be there.”

Understanding Video Compression

Video compression is the process of reducing the size of a video file by removing redundant or unnecessary data. This is done by encoding the video using a codec, which stands for “coder-decoder”. Codecs use complex algorithms to analyze each frame of a video and compress it in a way that minimizes the loss of quality.

There are two types of video compression: lossless and lossy. Lossless compression reduces the size of a video file without any loss of quality, but it’s not as effective as lossy compression in terms of file size reduction. Lossy compression, on the other hand, sacrifices some quality to achieve a smaller file size. The level of quality loss depends on the amount of compression applied.

When it comes to video compression, there are many factors to consider, including the resolution, bit rate, and frame rate. By adjusting these parameters, you can find the right balance between file size and quality for your particular needs.

As filmmaker and author Robert Rodriguez once said:

“Filmmaking is a chance to live many lifetimes.”

Compression Techniques for Audio and Video

There are many compression techniques used in audio and video, each with its own strengths and weaknesses. In audio, the most common type of compression is called “peak compression”, which reduces the volume of loud sounds that exceed a certain threshold. Another type of compression, called “multi-band compression”, divides the audio signal into multiple frequency bands and applies compression to each band separately.

For video compression, the most popular codecs are H.264 and HEVC (High-Efficiency Video Coding). H.264 is widely used for streaming video on the internet, while HEVC is more efficient but requires more processing

Free Download Mp4Gain

Mp4Gain Main Window

Mp4Gain Features

Free Download Mp4Gain

Audio Compression Techniques: Understanding the Basics

What is Audio Compression?

Audio compression is the process of reducing the size of digital audio files by removing redundant or unnecessary information, while maintaining the perceived quality of the original sound. This is done by using various algorithms that analyze and modify the audio data in a way that reduces its file size.

Types of Audio Compression Techniques

There are two main types of audio compression techniques: lossy and lossless.

Lossy Compression

Lossy compression algorithms are used to achieve high compression rates, but at the cost of some loss in quality. In lossy compression, some of the original audio data is discarded or modified in a way that reduces its size. The amount of data that is removed or modified depends on the compression algorithm used.

Some popular lossy compression algorithms include MP3, AAC, and WMA. These algorithms are commonly used for music streaming, online radio, and other applications where high compression rates are necessary.

Lossless Compression

Lossless compression algorithms are used to compress digital audio files without losing any information. These algorithms are designed to reduce the size of the file by removing redundancies in the data, but without modifying any of the original information.

Some popular lossless compression algorithms include FLAC, ALAC, and WAV. These algorithms are commonly used for high-quality music streaming and for archiving music collections.

How Audio Compression Works

Audio compression works by analyzing the original audio data and then modifying it in a way that reduces its size while maintaining its quality. This is done using various mathematical algorithms that compress the data.

The most common way to compress audio data is to use perceptual coding. This method takes advantage of the human ear’s limitations in hearing certain frequencies and sounds. By removing these sounds, the audio data can be compressed without the listener noticing any loss in quality.

Another method of audio compression is predictive coding. This method uses mathematical algorithms to predict the next sample in a waveform based on previous samples. The difference between the predicted sample and the actual sample is then compressed and stored.

Why Audio Compression is Important

Audio compression is important because it allows us to store and transmit audio data more efficiently. This means that we can store more audio files on our devices and transmit audio data faster over the internet. Without audio compression, it would be impossible to stream music or podcasts over the internet.

12 Common Questions About Audio Compression Techniques

1. What is the difference between lossy and lossless audio compression?

Lossy compression algorithms are designed to achieve high compression rates at the cost of some loss in quality, while lossless compression algorithms are designed to compress audio files without losing any information.

2. Which audio compression algorithm should I use?

The choice of audio compression algorithm depends on the intended use of the audio file. Lossy compression algorithms like MP3 and AAC are commonly used for music streaming and online radio, while lossless compression algorithms like FLAC and ALAC are commonly used for high-quality music streaming and archiving.

3. How much does audio compression affect the quality of the original sound?

The amount of quality loss in audio compression depends on the compression algorithm used and the degree of compression applied. Lossy compression algorithms generally result in some loss in quality, while lossless compression algorithms do not.

4. How can I tell if an audio file has been compressed?

You can usually tell if an audio file has been compressed by looking at its file extension. Lossy compressed files usually have extensions like MP3, AAC

The compression algorithm of an Mp3.

In addition to the physiological structural properties of the human ear, the function of the brain also plays a very important role.

Mp3 compression algorithm

The pitch in the sound is determined by the fundamental tone, while the timbre is determined by the harmonics, and the human brain will automatically complete the fundamental tone, even if the fundamental tone does not exist. For example, the bandwidth of a telephone is only 300~3200 Hz, but when we listen to a man with a base tone of 120 Hz talking on the telephone, we can still hear his correct tone and will not confuse a boy with a girl. . .

We still don’t know how the brain uses complex calculations to reconstruct this non-existent tone.

PS Add a little visual easter egg, can you see what’s weird about this image?

(Please read the answer to the end)

…………………………………………………………………………………………………………………… ………… ……… …………………………………………………………………………………………………………………… ………………………………………………………………………………………………………………………………… ………………………………………………………………………………………………………………………………… ……………………………………………………………………………………………………………………………… ……………………………………………………………………………………………………………………………… ……………………………
_ Your vision~ amazing! The human body still has too many unknown magic eggs waiting to be excavated~~

The compression algorithm of an Mp3.

The birth of the MP3 compression algorithm is nonsense of human organs in the digital age. The whole algorithm is not improved around the math, but rather optimized around how to fool the human hearing organ.

Mp3 compression algorithm

So this algorithm is very curious, Baidu finally found information after a long time, and has a little understanding of the principle of it, so please record it.

basic principle
There is a special effect of shading effect on the human hearing model.
The role of the cochlea is as a spectrum analyzer, converting sound waves into signals of different frequencies. The villous cells at each specific location will be stimulated by a specific frequency, but when the basilar membrane leads to fluctuations, the villous cells around it will also be stimulated. That is, if there is a frequency with a high volume, and at the same time there is a relatively weak frequency near it, the sound of the relatively weak frequency will be covered by the relatively loud sound, and our human ears have no way to distinguish the sound There is another sound of a weaker frequency.

To the human ear, the perception characteristics of sound do not change on a linear frequency scale (human hearing is not that good), but can be expressed in a series of limited frequency bands called critical frequency bands. Simply put, the entire frequency band is divided into several segments, and in each frequency band the auditory perception of the human ear is the same, that is, the psychoacoustic characteristics are the same.
Then, according to this principle, the mp3 compression work can be simply divided into two parts:

The first step: dividing the original audio data into several subcritical frequency bands according to certain principles;

Step 2: Analyze the frequency spectrum according to the psychoacoustic model to find the masking effect curve. Then, according to this curve, each sub-frequency band is quantized separately, and finally the compression of the audio is below the masking effect curve.

In this way, mp3 compression is done. And it is surprising that mp3 is really compressed in the digital world, but it belongs to compression without distortion for human perception.

Compression audio encoding Part 3

Compression audio encoding Part 3

I often hear what is called Hi-Res Audio. The sampling frequency is said to be 96 kHz or 192 kHz, which is over 48 kHz, the number of quantization bits is 24 bits, and the limit (high range) of human hearing is about 20 kHz, but it expresses frequencies higher than that. It will be. It is the same bit rate as the image from a long time ago. .. ..
By the way, it seems that dogs can hear up to 60 kHz and cats up to about 64 kHz.

Hi-res audio example
Sampling frequency Number of quantization bits Number of channels bit rate Frequency that can be expressed
192 kHz twenty-four 2 9.216 kbps 96 kHz
192 kHz 16 2 6,144 kbps 96 kHz
96 kHz twenty-four 2 4.608 kbps 48 kHz
96 kHz 16 2 3,072 kbps 48 kHz
48 kHz twenty-four 2 2,304 kbps 24 kHz
Considering the limit of human hearing (about 20 kHz), according to the sampling theorem, 48 kHz or 44.1 kHz is a sufficient frequency, but what about all of them? .. ..
In my case, I cannot distinguish the high resolution range, but it should be able to reproduce the discarded frequency at 48 kHz to 96 kHz, and when the number of quantization bits is in the 24-bit range, the sound pressure (dB) is a bit. Feels like I’m going up (?) (It’s just a story from my ears).
I’d like to make a comparison if I get the chance, but I don’t think I can tell by ear without a proper regenerator (like an expensive analog amp).

Is it time for cats and dogs to get verified in the acoustic industry? .. ..

Compression audio encoding Part 2

16-bit monaural PCM bit rate (for audio) (example)

Sampling frequency Number of quantization bits Number of channels bit rate Comments
32 kHz 16 1 512 kbps Super Wide Band
24 kHz 16 1 384 kbps
16 kHz 16 1 256 kbps Broadband
8 kHz 16 1 128 kbps Narrowband

Sampling rate
If you check the web, there are explanations like the sampling required to convert analog waveforms to digital conversion. For example, it shows how many samples of an audio signal input from a microphone are taken per second and digitized. The larger the sample, the greater the range that can be recorded. When an analog waveform is digitized, the frequency that can be expressed is half the sampling frequency (sampling theorem). For example, with a sampling frequency of 48 kHz, it can be expressed up to 24 kHz. At 8 kHz (narrow band) and 16 kHz (wide band), which are often used for audio, you can only hear up to 4 kHz and 8 kHz, respectively. The higher the sample rate, the higher the bit rate.

Sampling theorem
It is a very simple explanation, but it can express up to half the sample rate. When sampling a signal, if the interval is small, it can be restored close to the original signal, but if it is too thick, it cannot be restored (I would like to write a little more detail when I talk about signal processing or other time ).

44.1 kHz
Why is there a poorly separated rate of 44.1? .. ..
Isn’t the technician deliberately wearing an annoying watch to prevent music CDs from being easily copied? I heard something like that. When I searched, it seems this happened (?) Due to the convenience of an old PCM recorder. In this age, it is difficult to know what 44.1 kHz is in development. The 44.1 kHz ↔ 48 kHz sampling conversion is a headache. For example, USB audio (USB audio device class) exchanges data at 1 ms intervals. In the case of 48kHz, the data is 48 samples, but when considering 44.1kHz, it will be 44 samples (x9) and 45 samples (x1) in 10ms. If a sample of 45 samples is misled (tentatively), it will be 44.0kHz. I think it’s more like that with voice and music, and the human ear is mostly misleading (just my personal opinion).
However, the objective evaluation method will soon come to an end. For example, you can clearly see that you were fooled by a sine wave (sine wave) (maybe you are unexpectedly on the market).

Number of quantization bits
Sampling had to take a value in the direction of time (discretization), but quantization had to take a value in the direction of amplitude. The range that is possible to display the volume of the sound, which is heard often, “dynamic range 96 dB” means that the number of quantization bits is 16 bits and the music signal is played in the range of 0 to 65535 I can do it. The number of quantization bits is also called the bit depth or bit depth.

Bitrate
In communication, it indicates how many bits of data are transferred per hour and is generally expressed in bps (bit / s) of how many bits are transferred (processed) per second. If it is small, the size when saving as a file is small and there is space on the transmission line for communication. For example, when an audio (1 channel) is compressed to 1/3, the 3 channel audio can be sent at the same bit rate. Excuse the old story, but considering from the age of analog communication (analog mobile phone), digitization + compression will be able to support multiple calls with the same radio wave.

compressing using audio encoding

When compressing using audio encoding (AAC, MP3, etc.), the compression rate is determined by the bit rate at the time of encoding.

Specifically, if you set a low bitrate, the compression rate will be high and the file size when saved will be small, but what is the bitrate for the original sound source (PCM) without compression in the first place?

If you save it as PCM, the sound quality of the original sound will be obtained, but it can be a little inconvenient to save it without worrying about the file size. Also, depending on the application, I think the original sound size has enough memory capacity and the communication speed is correct. Therefore, I would like to write about the sample rate and bit rate that are often heard in digital audio.

The bit rate of digital audio is determined by the sample rate, the number of bits assigned to a sample (number of quantization bits), and the number of channels (stereo, monaural, etc.).

PCM bit rate (uncompressed) = sample rate x number of quantization bits x number of channels
As I wrote a bit last time, file containers like wav and mp4 format have this information as the header, so the application can see the header and play it. The compression rate of the encoding is determined by the bit rate specified at the time of encoding for this PCM (uncompressed) bit rate.
For example, as many of you know about music CDs, with 44.1 kHz stereo, this is the next bit rate.

Music CD bit rate: 44100Hz x 16bit x 2ch (stereo) = 1411.2kbps
When encoding this with MP3, AAC, etc., you will naturally specify a bitrate less than 1,411.2 kbps. For example, when encoding at 256 kbps, the compression rate is approximately 18% and the file size is 1/5 or less, assuming the original sound is 100%.

Encode 256 kbps music CDs: 256 kbps / 1,411.2 kbps = approximately 18%
Generally, the sample rates of audio devices actually connected to a PC are 48 kHz and 44.1 kHz for music, 16 kHz and 8 kHz for voice, such as microphones and headphones, and 32 kHz, 24 kHz, 22.05 kHz, etc.

The bit rate of PCM (uncompressed sound source) with 16-bit quantization bits is as follows.

Stereo (for music) PCM 16-bit bit rate (example)
Sampling frequency Number of quantization bits Number of channels bit rate Comments
48 kHz 16 2 1,536 kbps
44.1 kHz 16 2 1,411.2 kbps Music CD
32 kHz 16 2 1,024 kbps
24 kHz 16 2 768 kbps
22.05 kHz 16 2 705.6 kbps