MP3 file format is the so-called " open format"Supported by most manufacturers.
MP3 format One of the most common digital sound encoding formats. A feature of the sound coding in MP3 format is coding with losses. However, the coding is based on a special model, which takes into account the characteristics of the auditory perception. Therefore, the presence of losses does not lead to the catastrophic degradation of sound.
MP3 format files have become an actual standard, their playback is maintained by most popular operating systems, many CD players and DVD players and other devices.
Interestingly, the standard describes the storage format itself, and not a way of encoding sound. Due to this, there is a huge amount of means that serve to play sound in MP3 format.
Special codecs are used to encode audio in MP3 format.
Audio codec can belong to one of two types - hardware codec and software.
Hardware coding is performed using special microcircuits.
Software coding is performed using special computer programs.
The sound quality in MP3 format (with other things being equal) depends on the degree of compression (read from the amount of losses) and from the coding program. That is why branded players using codecs and processing systems sound signal From the well-known brands, normal devices collected from typical nodes are significantly superior to the quality of playback.
The quality of playback actually depends on the value of the data flow from the carrier. Sometimes the value of the data stream is called the flow width. There is a special term - Bitrate. The data flow rate is determined in kilobits per second and is indicated by KBS, KBPS, KB / S. The recording can be encoded in several ways - with a constant bitter and variable bitrate. A variable bit rate helps keep parts by increasing the data volume.
For quality playback of music Not all the speed of data flow see Paint 1
Data flow rates in MP3 format and scope
Table 1
The data given in Table 1 can only serve as a guideline. The fact is that at the time of the format of MP3, the quality of the audio instrument of mass demand was not very high. Many authoritative editions seriously argued that a 128 kb / s data stream is enough to high-quality sound playback.
Currently, a bit rate of at least 192 kb / s is considered quality. Moreover, the wide distribution of Hi-Fi, Hi-End and home theater systems led to a massive transition to high-quality sound playback.
Therefore, the ocrets of sound reproduction, imperceptible in the budget equipment of the past, become noticeable to "unprepared listener" using modern high-quality equipment. By the way, the level of this very "unprepared listener" has grown significantly.
In general, the idea of \u200b\u200bcompression (and especially compression with losses) gradually makes himself up. Appearing in the epoch expensive carriers of information and malaya bandwidth Data transmission channels, the idea of \u200b\u200bdata compression perfectly coped with its main task. However, gradually sound lovers are moving to higher bitrates (on compression with smaller losses), and even at all on the compression formats "without loss" or even without compression.
The practicality of compressed formats, and MP3 format in particular, led to the release of compact MP3 players arranged on memory chips or on miniature hard drives.
When choosing a model of a similar player, a question arises related to the amount of its memory. Naturally, the user wants to evaluate the number of musical material in advance, which he can save on its MP3 player.
Approximate data on the volume of files and the duration of the sound is collected in Table 2. When using Table 2, it is necessary to take into account that this is an exemplary data that allows you to estimate the required amount of memory of players or interchangeable media.
The duration of the sound of MP3 files and compression ratio
table 2
Bitrate |
1 minute recording |
Standard |
Standard |
Standard |
Note to Table 2
The high degree of compression corresponds to the value of 56 kb / s, the low degree of compression and high sound quality corresponds to 320 kb / s
Table 3 presents approximate data on the total duration of music records - player sounding time with a different memory.
Total MP3 Player Time Depending on Memory Volume
Table 3.
Sound duration |
Memory size, |
Bitrate, KB / S |
||||
Minutes |
||||||
Minutes |
||||||
Minutes |
||||||
Minutes |
||||||
Minutes |
||||||
Minutes |
||||||
As far as can be judged by Table 3, an 8 GB volume is enough to maintain records in the highest quality MP3 format in an amount suitable for listening to 8 hours every day during the week (7 days). Without repeats! Hardly someone really has a similar need.
Even if it is so, then updating the recordings on the player are not more often once a week.
2013 site. All Rights Reserved.
See you in the network!
How to enable Wi-Fi on ASUS laptop |
Over the past few years, it became a terrible fashionable and popular MP3 format. On any tray selling computer CDs, you can easily find more than a dozen disks of the "complete anthology of the group XXX", and below the modest such inscription - mp3. Most often for the full picture on the covers there is a fashionable phrase CD Quality - then you mean the quality, like Audio-CD. It is about this that will not only be next our story - about MP3, what they happen about the quality of sound in mp3.
About mp3 format
To begin with, we will understand a little with the subject area. What does this represent this mp3 in general?
Mp3, more correct name MPEG-1 Layer 3 is a standard for compressing audio information with losses. At the same time, the main purpose of creating the standard was to ensure the maximum "identical" source sound, as well as minimizing the volume of stored data. For this was created original scheme Coding - At the first stage, the digitized sound is divided into frequency components that pass through a series of filters.
The main difference between MP3 from previously existing standards is in filtration. Standard Developers created the so-called psychoacoustic model - a model that takes into account some features of human hearing, and on the basis of this model from the audio signal, those frequencies are filtered out, the absence of which hearing almost does not notice. At the second stage, the resulting stream is encoded according to the Huffman algorithm with a static table. The result and will be a mp3 stream.
In addition, ID3 tags can also be added to the MP3 file (tags containing the name of the song, performer, other information) and various service information.
Compression and bitrate modes
The width of the stream - the bit rate determines how many bits are necessary for encoding 1 second of music. The MP3 standard regulates the streams from 8kbit / S to 320Kbit / s. The most typical bit rate is 128kbit / s.
Based on the stream, it is easy to calculate how much one minute of music will occupy - you need to divide the bit at 8 (the number of bits in the bate) and multiply by 60 (seconds per minute) - we get the number of kilobytes. For the already mentioned flux of 128Kbit / S it will be 128/8 * 60 \u003d 960 kilobytes or near megabytes per minute of recording.
It is quite natural that the greater the bit rate, the more the sound details can be saved, the more realistic it sounds. In the selection of the bitrate when coding, you have to sacrifice any quality in favor of small size, or the size in favor of quality.
The easiest MP3 compression mode is a constant bitter mode (CBR, Constant Bitrate). Earlier, at almost 100% mp3 assemblies, a 128Kbit / S bit rate was used above, and the CD Quality inscription was present on the disks. Frankly, it's just a slightening lie. In practice, to distinguish the sound of such MP3 from the sound of the audio CD, it is impossible not just on the cheapest acoustics.
Quality level on a bit rate of 128Kbit / s is an approximately the level of the middle tape recorder on not the fresh film can be slightly better. You can also add that this bitrate is widespread in the entries available on the Internet.
To simplify the parsing of higher bitrates, I will write their mesh: 128kbit / s, 160kbit / s, 192kbit / s, 224kbit / s, 256kbit / s, 320kbit / s. So, the bitrates 160 and 192kbit / s are already noticeably better in quality than 128kbit / s, but the received files are still not so high. "Artifacts" (flaws) codec is almost imperceptible (at least on my system).
With a Bitret of 224, I never had to meet in my pure form, so I can't say anything about his quality, but it should be higher than on the previous step of the ladder of bitrates. In addition, I did not meet reviews that cover this bitrate. Apparently it is somehow due to the fact that the first bit rate of 256kbit / s is recognized in terms of the accuracy of the sound transmission, almost a complete lack of distortion. In the instructions for the LAME codec, this bitrate is even named as Studio Quality. And the ceiling - 320kbit / s is designed for those whose quality is more expensive, or for owners of very high-quality Hi-Fi or even Hi-End equipment.
We now turn to a slightly more complex issue - the variable bitrate mode (VBR, VARIABLE BITRATE). Here the concept of bitrate is very blurry, the codecs "for the user" generally use adjustment only in quality (such as Xing Audio Catalyst). Other (lame) allow you to ask extra options - Minimal and maximum bitrates, again quality.
When encoding the VBR codec, it selects the desired bitrate itself, based on the parameters given to it, and during the encoded fragment, the bitrate may change. An already mentioned psychoacoustic model is used to assess the desired bitrate. However, the model (since it is not perfect, nothing in our world is perfect) sometimes shows incorrect results. This leads to inclusion of the assessment, and, accordingly, the fall of actually audible sound quality.
The developers of the LAME codec are advised in this case to set the minimum bitrate threshold to avoid very bad results. The varieties of VBR refers and encoding ABR (Average Bitrate), averaged bitrate. IN lately Only positive responses about this mode are heard in the reviews, especially ABR on 256kbit / s. This mode works almost as well as VBR, with the exception that the codec holds the average specified value. IN currently I am known to me only one codec having the ABR mode is LAME.
Choice Codec
Literally recently, the user who wanted to get a decent quality mp3 was not a very large selection - it is some ISO-based codec (based on the codec codec codec released by International Standarts Organization), or codec from IIS Fraunhofer (Institute - MP3 Developer ). Plus codecs in Xing products.
Honish different reviews, and making small own research, I came to the conclusion about the branch of the firm of the firm Xing - this ... they are better not to use. Even in relatively new versions, all their products that can create mp3 built-in means do it as much as possible.
There is also a lot of "pioneering" crafts, elbowed on a stovered in the Xing codec (almost all contain a Tompg.exe file as part). For a long time Their main advantage was speed (to the detriment of quality), but today the LAME codec shows comparable speed with higher quality. In addition, Xing products generally speaking costs money, while Lame is free by definition.
Next, I will go on IIS Fraunhofer products. All their programs for compression mp3, available for free, are highly cut down by the possibilities of the versions of their commercial products. Then, all of their codecs did not develop for a long time, and do not contain new tools, supporting VBR / ABR, in addition, not differing in special speed. The only justified application - compression on bitrates below 128kbit / s - they carried out special optimization for low bitrates (places, however, with a violation of the standard).
Different codecs based on ISO code suffer in principle with the same disadvantage - low-quality compression on bitrates below 192kbit / s. In addition, most of them (including bladeenc) are pretty slow.
In my opinion, the most optimal option is the LAME codec. Started as a free codec based on the ISO code, during the development process it has grown and now all reviews when comparing MP3s with other formats are used precisely as a reference for MP3. Little more than a year ago, the LAME project finally got rid of the ISO code and can now be considered a completely independent codec.
The development of the codec is quite intensive, it is constantly updated, correct errors. In addition, it is possible to use LAME not only under Windows, but also for various options for UNIX systems, it also works in pure DOS. Again, completely free, available source (For lovers in it, you dig in), from multiple sites already compiled binary files (.exe and.dll), optimized for different processors available.
There is also a slightly trimmed version of the LAME - Gogo-No-Coda encoder, which shows fantastic results (twice as fast as fast LAME).
So what is the bitrate and what mode to use?
Considering all of the above, I would recommend putting a MP3 file either with a 320kbit / s stream, CBR mode, or 256kbit / S, ABR. The first in my opinion is somewhat preferable, because You get the most accessible quality within the format. For recordings to "listen and erase" a couple of times, it is reasonable to use ABR 192Kbit / s.
And one more - it is better not to use the bitrate for some long storage below 192kbit / s - if only the record with which MP3 was made, you are not constantly at hand (although remember that the analog record on the magnetic tape is deteriorated over time) .
Very often, the argument that I hear in favor of low bitrates and the "curve" of compression is "I have bad acoustics, and I still can't hear the difference." Everything can change, or you have to use your archive on a decent equipment, and it will not be possible to get to the initial record. The answer is absolutely not thorough, I can bring the case from my own practice.
In our city Pavlovo was once a small club, where the music was played from the computer (MP3 with a bit rate not higher than 160Kbit / s). The club further passed away, and the computer with music archives moved to another firm engaged in mass events. Imagine that they took to twist this music at the bottom of the city! Horror, when all defects brought by packaging on such a small bit rate were heard on more or less decent acoustics. The sound was worse than with their seaside tape recorder with semi-advanced cassettes. It would be reasonable to avoid the repeat of other people's mistakes, right?
Test equipment and software
Computer: Athlon TB 650MHz, M / B ACORP 7KTA 100MHz FSB, 128MB RAM PC-133, HDD Quantum 40GB 5400RPM, SoundBlaster 16 Vibra, AC97 Codec.
Audio system: Radiotehnika amplifier U-7111, a pair of Radiotehnika S-90B speakers.
By: Windows98 SE, Winamp 2.75, EAC 0.9PB11, LAME 3.90A, GOGO-NO-CODA 3.07A
The bitrate is indicated as one of the main characteristics of video and audio recordings. Most users got used to think that it defines the quality of the file being downloaded. But what is bitrates and how do they actually characterize music files and videos? Consider this in more detail.
What is bitrates?
Bitrate is a value that displays the number of information units (megabit or kilobit), which is fixed in one second of the file playback. Accordingly, it is measured in megabits per second (MBPS) or kilobits per second (KBPS). Otherwise, the bitrate can be described as a bandwidth bandwidth. This feature is important for those who want to convert files, because with the same duration, the greater bit rate will lead to an increase in the file. In addition to the size, the sound quality changes. Reducing the size when lowing the bitrate is called compression.
A common musical is an audio file, compressed to such an extent that the standard disk is placed up to 12 hours of music. At the same time, the quality remains high enough due to psychoacoustic compression: from the entire range, sounds with those frequencies and volumes of volume, which are not captured by the human ear are removed. Selected sounds are formed into separate blocks, called frames. Frames have the same sound duration and compress on a given algorithm. When the music is played, the signal is recreated from decoded blocks in a specific sequence.
What is commonly used compression?
The audio bit rate is most often 256 kbps. With this value, the audio recording is compressed in the amount of approximately 6 times, due to which one disc can be recorded 6 times more music than before compression. If the bitrate is reduced to 128 kbps, then one disk will fit already 12 times more music, but the sound quality will be noticeably lower. Music recorded as 128 kbit / s is most often offered for listening on the Internet, since in pursuit of increasing the page loading speed of the resource owners go to any sacrifices. Many users note that its quality is far from perfect.
Now, when it became clear what bitrates are, it's time to determine their optimal level. Both lovers and professionals are infinitely arguing how the bitrate affects the sound quality and does it affect at all. On musical albums, as a rule, a bitrate is indicated. The same disc recorded as 128 Kbps and 256 Kbps will vary by price twice.
Optimal bitrate under different listening conditions
For many people, twelve-fold compression does not represent any damage, while others argue that they cannot listen to music with bitrate lower than 320 kbps. Paradoxically, but those and others are right. The fact is that ultimately the quality of playback depends not on the reproduction conditions and even from the type of music.
For example, the song is played on the tape recorder installed in the domestic car. In this case, quality at 192 Kbps will be quite sufficient. A higher bitrate will improve the sound quality, but the difference will not be noticeable due to the high level of noise during the trip. If the music plays on home computer or a portable player, then not less than 256 kbps. If the signal is not subject to change, is transmitted to external devices and excluded on expensive imported columns, then it is necessary to resort to minimal compression if possible. It is possible when bitrate 320 kbps.
Optimal bitrate for various musical styles
Music with a high bit rate is not always needed. Popular music, as a rule, sounds quite good when bitty 192-256 kbps. It is possible to install a higher quality, but there is no point in this: the pop songs are short-lived, so the saving of the disk space should be priority. In addition, the quality of the source records is also mediocre, so the increase in the bitrate may not affect the quality of the playable file. For listening in transport and in unofficial parties, the average quality is enough.
If we are talking about classical music, works of legendary rock groups or rare copyright songs, then quality should be above all. When purchasing such music, you need to look at the bitrate specified on the disc packaging. If the song is loaded from the Internet, then such information should be present on the download page. In addition, the bitrate is displayed in the player during playback.
Bitrates video files
It was said above that such a bitrate of audio recordings. But what is a bitrate video? Given that the video is played as a sequence of sounds and images, the definition of the bitrate will be similar. The presence of the video detects the file, but ultimately the image for the processor is the same zeros and units as sounds. The principle of information encryption is the same for all types of files.
What is one of the most common and deeply rooted delusions in the world of music lovers?Save and read later - 
Approx. Perev.: This translation of the second (of four) parts of the Deployed Article Article of Christopher Montgomery (Creator OGG Free Software and Vorbis) is about what, in his opinion, is one of the most common and deeply rooted delusions in the world of music lovers.
The frequency of 192 kHz is considered harmful
Musical digital files with a frequency of 192 kHz do not bring any benefit, but still have some influence. In practice it turns out that their playback quality is slightly worse, and during playback, ultrasound waves arise.
And audio developers and power amplifiers are affected by distortion, and distortion, as a rule, quickly increase in high and low frequencies. If the same speaker reproduces the ultrasound along with frequencies from the audible range, then any nonlinear characteristic will shift the part of the ultrasonic range in the audio spectrum in the form of disordered uncontrolled nonlinear distortions covering the entire audible sound range. Nonlinearity in power amplifier will lead to the same effect. These effects are difficult to notice, but the tests confirmed that both types of distortion can be heard.
The graph above shows the distortions obtained as a result of the intermodulation of sound with a frequency of 30 kHz and 33 kHz in the theoretical amplifier with a constant nonlinear distortion coefficient (book) about 0.09%. Distortion is visible throughout the spectrum, even at lower frequencies.
Failure ultrasonic waves contribute to intermodulation distortions in the audible range (light blue zone). Systems that are not intended to reproduce ultrasound, usually have higher levels of distortion, about 20 kHz, further contributing to the intermodulation. Expansion of the frequency range To include ultrasound requires compromises that reduce the noise and the activity of distortion within the auditor, but in any case, unnecessary reproduction of the ultrasonic component will worsen the playback quality.
There are several ways to avoid additional distortions:
- The speaker intended only for the reproduction of ultrasound, the amplifier and the signal spectrum, to divide and independently play ultrasound, which you cannot hear that it does not affect other sounds.
- Amplifiers and converters designed to reproduce a wider frequency spectrum so that ultrasound does not cause audible nonlinear distortion. Due to the additional costs and difficulty of execution, additional frequency range will reduce playback quality in the hearing part of the spectrum.
- Qualitatively designed speakers and amplifiers that do not reproduce ultrasound at all.
- For a start, you can not encode such a wide range of frequencies. You cannot (and should not) hear ultrasound nonlinear distortion in the audible frequency band, if there is no ultrasonic component in it.
All these ways are aimed at solving one problem, but only 4 ways have some meaning.
If you are interested in the possibility of your own system, the following samples contain: sound with a frequency of 30 kHz and 33 kHz in a 24/96 WAV format, a longer version in FLAC format, several melodies and cutting of conventional songs with a frequency shown to 24 kHz so that They fully fall into the ultrasonic range from 24 kHz to 46 kHz.
Tests for measuring nonlinear distortion:
- Sound 30 kHz + sound 33 kHz (24 bits / 96 kHz)
- 26 kHz melodies - 48 kHz (24 bits / 96 kHz)
- 26 kHz melodies - 96 kHz (24 bits / 192 kHz)
- Cutting from songs shown in 24 kHz (24 bits / 96 kHz WAV) (original version of cutting) (16 bits / 44.1 kHz WAV)
Suppose your system can reproduce all formats with 96 kHz sampling frequencies. When playing the above files, you should not hear anything, no noise, no whistle, no clicks or what other sounds. If you hear something, then your system has nonlinear characteristic and causes audible nonlinear ultrasound distortions. Be careful when increasing the volume, if you get into the zone of digital or analog signal level limit, even soft, then this can cause a loud intermodulation noise.
In general, it is not a fact that non-linear distortions from ultrasound will be heard on a specific system. The accuracy of distortion can be both insignificant and quite noticeable. In any case, the ultrasonic component is never worthy, and in a variety of audio systems will lead to a strong reduction in sound reproduction quality. In systems that it does not harm, the ability to process ultrasound can be saved, but you can use the resource instead to improve the sound quality of the audible range.
Misunderstanding the discretization process
The theory of sampling is often incomprehensible without the context of signal processing. And it is not surprising that most people, even ingenious doctors of science in other areas, usually do not understand her. It is also not surprising that many people do not even realize that they understand it wrong.
Discretized signals are often depicted as an uneven ladder, as in the figure above (red), which looks like a rough approximation to the original signal. However, such a representation is mathematically accurate, and when there is a conversion to an analog signal, its graph becomes smooth (blue line in the figure).
The most common misconception is that, allegedly, sampling is rude and leads to loss of information. The discrete signal is often depicted as a toothed, angular stepped copy of the original perfectly smooth wave. If you think so, you can assume that the greater the frequency of sampling (and the greater the bit on the count), the smaller the steps and the more accurate will be the approximation. The digital signal will increasingly resemble an analog form until it takes it to the form at a sampling frequency that tends to infinity.
By analogy, many people who are not related to digital signal processing, looking at the image below, will say: "Fu!" It may seem that the discrete signal does not represent high frequencies of the analog wave, or, in other words, with increasing sound frequency, the sampling quality drops, and the frequency response is worsening or becomes sensitive to the phase input signal.
It just looks like this. These beliefs are incorrect!
Comment from 04.04.2013: As an answer to all mail, regarding digital signals and steps I received, I will show the real behavior of the digital signal on real equipment in our Digital Show & Tell video, so you can not believe me for word.
All signals with a frequency below the Nyquist frequency (half of the sampling frequency) during the sampling will be captured perfectly and completely, and the infinitely high frequency of sampling is not needed for this. Discretization does not affect the frequency response or phase. Analog signal can be restored without loss - the same smooth and synchronous as the original one.
Will you argue with mathematics, but what's the difficulty? The most famous is the requirement of the limitation of the strip. Signals with frequencies above the Nyquist frequency must be filtered before sampling to avoid distortion due to the imposition of spectra. The role of this filter is the infamous smoothing filter. Suppression of discretization interference, in practice, cannot pass ideal, but modern technologies allow you to approach the ideal result very close. And we came to overpressure.
Excessive discretization
Discretization frequencies Over 48 kHz are not related to high accuracy of audio reproduction, but they are necessary for some modern technologies. Excessive discretization (oversampling) is the most significant of them.
The idea of \u200b\u200boversampling is simple and elegant. You can remember from my video "Digital Multimedia. Handbook for beginner gickers "that high frequencies of sampling provide a much greater gap between the highest frequency, which concerns us (20 kHz) and the Nyquist frequency (half of the sampling frequency). This allows you to use simpler and more reliable smoothing filters and increase the accuracy of playback. This is an additional space between 20 kHz and Nyquist frequency, essentially simply shock absorber for an analog filter.
The figure above shows diagrams from the video "Digital Multimedia. Handbook for beginner guys "illustrating the transition bandwidth for DAC or ADC at a frequency of 48 kHz (left) and 96 kHz (right).
It's only half of the case because digital filters Have less practical restrictions in contrast to analog, and we can complete smoothing with greater accuracy and efficiency. The high-frequency raw signal passes through a digital smoothing filter, which does not experience problems with the placement of the filter transition band in a limited space. After the smoothing is completed, additional discrete segments in the shock absorbing space simply lean. Playback of the reversed signal passes in the reverse order.
This means that signals with a low sampling rate (44.1 kHz or 48 kHz) can have the same playback accuracy, smoothness of ACH and low overlap, as signals with a 192 kHz sampling frequency or higher, but neither one of them will manifest Disadvantages (ultrasound waves, causing intermodulation distortion, increased file size). Almost all modern DACs and ADCs produce redundant discretion at very high speeds, and few know about it because it happens automatically inside the device.
DAC and ADCs did not always be able to over-encryptize. Thirty years ago, some sound recording consoles used high sampling frequencies for recording, using only analog filters. This high-frequency signal was then used to create master disks. Digital smoothing and decimation (re-discretization with a lower frequency for CD and DAT) took place at the last stage of the recording. It could become one of the early reasons why the sampling frequencies of 96 kHz and 192 kHz became associated with the production of professional sound recordings.
16 bits against 24 bits
Ok, now we know that keeping music in 192 kHz format does not make sense. The topic is closed. But what about 16-bit and 24-bit audio? What is better?
The 16-bit audio with pulse-code modulation really does not completely cover the theoretical dynamic sound range, which is capable of hearing a person in ideal conditions. There is also (and always) the reasons to use more than 16 bits to record audio.
None of these reasons have a relation to reproduction of sound - in this situation, 24-bit audio is as useless, as well as discretization at 192 kHz. The good news is the fact that the use of 24-bit quantization does not harm the quality of the sound, but simply does not make it worse and takes up an excess place.
Notes to Part 2
6. Many of the systems that are unable to reproduce 96 kHz samples will not refuse to reproduce them, and they will imperceptibly subdidiscress them to the frequency of 48 kHz. In this case, the sound will not be played at all, and nothing will be on the record, regardless of the degree of nonlinearity of the system.
7. Redescript is not the only way to work with high sampling frequencies in signal processing. There are several theoretical methods to get a band-limited sound with a high sampling frequency and avoid decimation, even if it is later subdivided for writing on discs. It is not yet clear whether such methods are used in practice, since the development of most professional installations are kept secret.
8. It does not matter, historically it happened or not, but many specialists today use high permissionsbecause they mistakenly believe that the sound with the preserved content outside of 20 kHz sounds better. Just like consumers.
This article read 33 932 times
