Immediately mention that the article says ONLY about the general characteristics and will not include some details. In the future, the Lifehacker will conduct his own impartial research. And today we will try to generalize somehow known experience.
There is an analog and a figure.
Analog – good, but short-lived and uncomfortable. Therefore, analog carriers, despite the high sales of vinyl, will not return.
Audio can be of three main types:
- In a format that does not use compression;
- in a format using lossless compression;
- in a format that uses lossy compression.
At first glance, formats that use lossless compression are more promising. This is not always the case, as we will discuss in more detail in one of the following materials. Uncompressed formats do not make any sense, except for storing master records needed when creating audio content. They are easier to recover. To store and listen to home records unnecessary.
Of the many parameters of the user’s digital audio, the sampling rate (the accuracy of digitizing the analog signal over time), the bit depth (the accuracy of amplitude-quantization in loudness), the bit rate (the amount of information contained in the file in terms of a second) must first of all worry.
Today, let’s talk about lossy.
For a compressed sound, the concept of a psychoacoustic model is very important – the representation of scientists and engineers about how a person perceives sound. The ear perceives the entire spectrum of acoustic waves arriving at it. However, it processes the brain signals.
The reference value of a person’s audible range is from 16 Hz to 20 kHz, but simultaneously he can not hear and realize all incoming sounds.
The hearing is discrete, and its sound susceptibility is nonlinear.
Modern psychoacoustic models accurately assess the human hearing and are constantly improving. In fact, despite the assurances of music lovers, musicians and audiophiles, for the average untrained ear, the original form of MP3 in maximum quality became extremely perceptible. There are exceptions, they can not fail to be. But they are not always easily visible with blind listening.
Formats using psychoacoustic models of compression
There are quite a few formats for compressing audio with loss. The most common today are the following.
In general, a file with the extension *. Logg is a “container”: inside there can be several audio records with their own tags and characteristics. Most of the files stored in it are compressed by the Ogg Vorbis codec, although others, including MP3 or FLAC, can be used.
As its main advantages it is necessary to name a large range of possible parameters for encoding: the audio sampling frequency can reach 192 kHz, the bit depth is 32 bits. By default, OGG uses a variable bit rate (although this is not indicated when displaying properties), which can reach 1,000 kbps.
Unlike free OGG, MP3 is very important for modern acoustics by the Fraunhofer Society – an association of applied research institutes in Germany. In the environment of audiophiles, by the way, this is an extremely respected office, however, they do not like to admit it. But their developments are closely monitored.
Unlike OGG, it can have both a variable (VBR) and a constant bitrate (CBR). By the way, it was thanks to MP3 that it was found out that not every record can be qualitatively encoded with variable bit rate (for the reasons see above, the coding algorithms and their results in this case can be different when coding the same source).
In view of old age, MP3 has significant limitations: the bit depth can be 16-24 bits, the sampling rate is expressed only in discrete values (8, 11,025, 12, 16, 22,05, 24, 32, 44,1, 48), the bitrate is limited to 320 kbit / s. In addition, in the normal MP3 version, the number of channels is limited to two.
The same rake, only in profile. Also the development of the Fraunhofer Society. Later and uses a different psychoacoustic model, more modern. The public information allows to draw a conclusion: yes, they managed to improve their own creation.
Even if we rely on the simplest figures, AAC is a more flexible format. The size of the files obtained with this development is from 16 to 24, the sampling frequency also, if desired, will not lose the sound picture and lies in the range of 8-192 kHz. The data stream generally approaches that of lossless formats (up to 512 kbps), while the maximum number of AAC-file channels reaches 48.
Which format should be called unequivocally the best
If you consider that AAC – rethought after a decade of MP3, then the choice in his favor. If desired, it makes sense to compare only MP3 and OGG. Let’s look at the pictures made by the respected Andrei Aspidov from ixbt.com:
On the charts – a good AudioCD, compressed OGG with a variable bit rate of 350 kbit / s and MP3 using Lame. The lower the graph, the closer the sound to the original. It turns out a very interesting picture. Despite the fact that the MP3 has clearly cut off high frequencies, unlike OGG, which shows a blockage below 2 kHz.
The frequency-time distribution of sound indicates no less interesting things. At a constant bitrate of 320 kbps MP3 almost coincides with the original recording. It seems, now everything falls into place. But … In fact, everything is still more confused.
Why generally use lossy when there is an affordable lossless
The fact is that most analog records do not contain the amount of information that should be stored in high-quality formats. Do not forget that the native sampling frequency for the CD is 44.1 kHz, quantization is only 16 bits.
The previous graphs demonstrate well the high accuracy of MP3 transmission. But for audio cassettes, magnetic tape (if, of course, this is not master-teyp), the characteristics of AudioCD are unattainable. And for mass studio equipment, the ability to record analog audio, corresponding to AudioCD, appeared relatively recently. There is no point in digitizing in FLAC (and even more so in WAV) a concert record or a record from the pre-digital era, especially those made with magnetic media. They do not contain those spectra and the amount of information that containers can store without compression.
What has changed today?
A rare sound engineer makes a digital master record (with which then multiplication is performed on physical media) using modern technologies to the fullest. Therefore, the chance that the 24-bit track is actually only 16-bit is extremely high.
Analog high-quality recording on high-quality equipment is even more difficult to meet today – if only for fans of this sound. So, for example, is Jack White, the ex-leader of the band White Stripes. At the same time, some of his records refer to variations of lo-fi, and to look beyond the limits of the sound characteristics of the track becomes a kind of pleasure for gourmets.
If you imagine an ideal source, then only a trained hearing or listening on high-quality audio equipment will find a compressed file. And already relying on this (and not forgetting about perception), it is worth the following conclusion:
The AAC is necessary and sufficient for the average pricing equipment, in the absence of which (and in the absence of source codes that can be encoded in AAC) – MP3 with a constant bitrate of 320 kbit / s, created with the codec Lame 3.93 (recommended keys for decoding: -cbr -b320 -q0 -k -ms).
The exception is the records originally received in high quality, say, recorded on DVD-Audio, SACD, or records originally collected in DSD (or similar format) with a high bit rate.
Although there are some features of lossless. And we will tell about them next time.
The author does not like Apple. The author appreciates the achievements of “Fraunhofer” very much and was extremely surprised to learn that AAC is their work. 🙂