The times when a personal computer communicated with the user exclusively through the built-in “beeper” are, fortunately, long gone. But if you are used to listening to music, and not a set of sounds, the standard PC capabilities are unlikely to suit you. Why? It's simple. Organization based personal computer A high-tech music center presupposes that the user has not only certain skills and knowledge, but also some material resources. And only irresponsible optimists, far from real life, can count on the fact that for 10–20 dollars you will get a branch of La Scala in your apartment.

What should those music lovers do who, even when working at a computer, are not ready to give up their favorite tunes, because not everyone can afford to spend several tens of thousands of dollars? Believe me, there is a way out. Perhaps a compromise, half-hearted and not at all ideal, but at least real. Perfectionists and refined audiophiles will certainly remain dissatisfied, but less pathetic users will find the proposed scheme quite acceptable.

Music lovers listen to it everywhere and always

What will you have to give up?

As we have already found out, “sounds” and “music” are exactly the same thing.. Almost any computer can provide the first, but providing the second is much more difficult, and it cannot be done without some sacrifices. So, what will go under the knife?

What do you need to be prepared for?

You will have to seriously weigh your own financial capabilities, since the organization high-quality sound on a PC - the event is in any case quite expensive. You can try to avoid large expenses, but if you have 50–100 dollars, it’s better to forget about music on your computer right away. The conditional “entry threshold” is twice as high, and this will be quite a compromise option with many reservations.

Let's go to the store

So, have you weighed the pros and cons and are mentally resigned to the need to spend some money? Then it’s worth figuring out how to do this most effectively.

1. Sound card. Creative, a very popular company among PC users, has been churning out outright consumer goods for several years now, which is very far from the legendary X-Fi Elite Pro, so it makes sense to take a closer look at sound cards from Asus. Essence STX II will cost 6,500–7,000 rubles, but this will be a purchase for many years to come. The first revision of this board will cost slightly less, and in case of strict budget savings, you can choose it. You can, of course, save even more and agree to Xonar DSX, but there is a considerable risk that very soon you will “get the taste” and begin to regret the purchase. But audiophile boards costing $300–800 require a much more thoughtful approach, which is why they are far from the best choice for getting to know the world of computer music. The main condition is a high-quality elementary base and the presence of a separate headphone amplifier.

Legendary X-Fi Elite Pro sound card

1. Playback device (speaker systems). As we have already found out, it is better to forget about plastic Genius speakers or crafts from Sven right away. Budget models from the same Edifier (R1280T, R1800) or Microlab (Solo 7c, Solo 15) will suit you. Of course, you can look at the budget Hi-Fi segment, but only if you can spend much more (up to $1000).
2. Playback device (headphones). Not a bad option if you don’t like the prospect of dealing with bulky and uncomfortable speakers (or if your living conditions don’t allow for it). Mid-budget models AKG (512mk2, ~1600 RUR), Audio-technica (ATH-M20X, ~3500 RUR) or JVC (HA-RX700, almost 4000 RUR) will cost slightly less than full-size acoustics at best quality sound But you should understand that not everyone likes wearing headphones for long periods of time.

Music library update

We will not consider the issue of the legality of downloading music from the Internet, but you will have to refuse regular mp3s with a bitrate of 128 kbit. You can find high-quality audio CD rips on the Internet without any problems (flac, ape, wv formats), but their size - approximately 300-400 MB - will make many people think about purchasing an additional hard drive.

Software player

Winamp, which has faithfully served many generations of users, or its clones like AIMP are poorly suited for serious listening to music. Therefore, we strongly advise you to get something more solid, convenient and functional (Foobar, MusicBee). These players have powerful cataloging tools and are capable of outputting sound bypassing the Windows mixer (WASAPI and kernel streaming modes). The only problem is the initial setup required.

MusicBee - a super bee for music lovers

Acoustics placement

The accepted layout of the speakers (at the edges of the table on which the computer stands) creates a very narrow stereo panorama. This cannot be called an absolute drawback, and many music lovers even like this sound delivery. But if you like classical music concerts or performances of rock bands, it is still better to space the acoustics as far as possible (be prepared to buy 10–20 meters of acoustic cable).

Preparing the system unit and OS

Let's be honest, this is not the most important point, since the likelihood of interference and interference occurring inside the computer is very small (at least if you do not suffer from advanced audiophilia). But if the power supply is working at its limit, and Windows is “bending” due to numerous errors in the registry, do not be surprised if instead of music you hear wheezing or “white noise”.

The post will be useful, if not for music lovers, then at least for people who didn’t have a bear step on their ears in childhood... Also for those who are on Windows 7 (if anyone doesn’t know, the sound quality in 7 (and Vista too) has suffered sooooo much compared to the same Windows XP). By the way, why this happened is also described below. Go...

The main criterion for playback quality is the amount of distortion that is introduced into the audio stream when passing through the audio path. Distortion, in turn, is inevitably introduced during any processing, so our main task is to minimize or completely eliminate sound processing in the section of the path along which the signal is transmitted digitally.

The sound path in our case has the following form:

1) Source- sound in MP3, FLAC, OGG, WAV etc. format.

2) Player- actually foobar2000. Contains:

• Decoder


• Sound handlers


• Conclusion

(see below for more details)

3) Windows mixer, driver sound card - very closely interacting elements. At this stage, software audio processing (often of poor quality) can be performed, which is extremely undesirable.

4) DSP sound card- this is already a hardware component. The Digital Signal Processor is located directly on the sound card and performs the necessary processing of the audio stream, as well as processing in accordance with the settings in the sound card mixer (most of the operations that this block performs depend on the settings of the sound card and the sound card model itself).

5) DAC- Digital-to-Analog Converter. Converts the received digital audio data into analog form for further amplification and supply to the analog output of the sound card (from it to headphones or an external amplifier).

1. Source

Undoubtedly, this link is the basis. If the recording is initially of poor quality, there is little point in wasting effort on setting up the rest of the path.

Sources can be divided into two types:

• Lossless(wav, flac, wv, ape, tak, tta, ofr etc.) - lossless encoded audio (when decoding, the resulting audio stream exactly matches what was encoded in Lossless - similar to unpacking files from a ZIP archive)


• Lossy(mp3, ogg, aac, mpc etc.) - lossy compressed audio. During encoding, irreversible losses of some part of the information occur (namely, that part of the sound data that a person does not perceive or almost does not perceive is cut out).

If in the first case the problems of high-quality playback are minimized, then in the second case, for the highest quality playback it is necessary to take certain measures (see “Recommendations for playing audio material”).

2. Software player foobar2000

Now let's take a closer look at everything related to foobar2000. One of the advantages of this player is a very thoughtful and transparent path. To clearly demonstrate this, I drew a block diagram (click to view):

Notes:
1. As can be seen from the diagram, the presence of all components is not necessary - some can be turned off, and some are turned on only when necessary. For example, a decoder is needed only for compressed formats, and a bit depth converter is turned on when the input data format does not match the output data format.
2. For Windows 7, floating-point output is possible (32-bit output via DS).

2.1 Decoder

Decodes the input data, resulting in an uncompressed audio stream in the form of pulse code modulation. Most lossy decoders operate in floating point mode (32-bit). The output of lossless decoders is data with parameters similar to the original audio (which was compressed).

2.2 Postprocessor

This element was included by the developer in the tract relatively recently. Acts as a decoder for HDCD and DTS formats (only lossless sources - for now FLAC and WV). Since the data from a conventional decoder must be transmitted bit-for-bit to successfully decode these formats, the postprocessor is located in the path immediately after it.

2.3 Handlers

Perform digital processing of the audio stream. It should be noted that they should be used only when necessary, because... Almost any sound processing - even changing the volume - will certainly introduce some distortion. All handlers operate in floating point mode by default. Foobar2000 digital processors include:

2.3.1 ReplayGain

First of all, it should be noted that any lossy audio file does not contain a digital audio stream as such. It contains its description using various functions, etc., from which you can restore approximate form source wave (which is what decoders do lossy formats). And so, when decoding, one nuance arises: because encoding occurs with losses (I repeat: the data in a lossy audio file allows you to restore only the approximate shape of the original wave), the output samples have a level different from what was at the input.

What does this threaten? The foobar2000 decoder operates in a floating point format, which allows it to process and describe a wave not only within a certain range (for example, 16 bits with a fixed point: 2^16 = 65536 possible level values). But the fact is that after the signal is transmitted by the player to output, it is automatically converted to a fixed-point format.

Let's look at an example. If we convert a sample with a level of 1.000000 (floating point) to a 16-bit fixed point format, we will get the number 65536 - this is the maximum level for 16-bit fixed point. But let's not forget that when decoding lossy, the original signal is restored approximately, and we can get samples with more high levels, for example 1.124325. And it will no longer be possible to correctly describe this value in fixed point mode, i.e. the level of this sample will automatically be equal to the maximum - 65535.

So what do we get? Thus, all sections of the wave (consisting of sample points) beyond the maximum level are “cut off”, as a result of which, instead of a sinusoid, for example, we get something similar to rectangular pulses (if the “tops” of the sinusoid are above the maximum, they will be “cut off” "), which means that we have additional nonlinear distortions (the level of distortion depends on what part of the sound wave goes beyond the maximum). Something similar can be heard when the speakers are “locked” - when you apply a signal to the speakers with a power exceeding the maximum permissible - the diffusers cannot move further maximum level deviations, the same distortions are obtained as in our case. The first phenomenon (with a digital audio stream) is software clipping, the second (with diffusers) is hardware clipping (in some other cases this phenomenon is called “overload”).

Why do we need ReplayGain in our case?

a) To determine whether the level of the decoded signal is outside the permissible limits, namely to determine the peak recording level - this is done by ReplayGain Scanner.

b) Lower the level of the track so that it fits within the acceptable range - after the scanner has scanned, it writes ReplayGain tags (with information about the peak level), and the player, when playing, reads these tags and lowers the volume level of the entire recording (namely, the entire - so as not to change the volume balance between individual sections) so that the peak is at the maximum level (0 dB on the peak meter), and everything else, of course, is not higher than this level.

Setting:

Source mode:track, if you want to maintain volume balance only within one track, album- if the volume balance between the tracks of the entire album is important to you.

Processing:Prevent clipping according to peak- the best option in most cases, which will only prevent clipping (the level at which the volume will be reduced will be calculated using the track peak or album peak tag - depending on the selected source mode).

Apply gain- in this mode, ReplayGain will help equalize the perceived loudness of the tracks you are listening to (often the perceived loudness is not related to peak levels, since it is determined by the method of psychoacoustic analysis); This option is not recommended, because in this case, the volume of the tracks can change greatly, which only worsens the situation in terms of quality.

Apply gain and prevent clipping according to peak. As a result of analyzing a track, the ReplayGain utility may consider it too quiet and write a positive gain value in the tags; the volume when playing such a track will be overestimated, as a result of which the peaks of the resulting signal may be higher than the maximum level. This mode allows the utility to analyze not only the track (or album - depending on the source mode) gain tag during playback, but also the track (album) peak tag and calculate the maximum allowable gain to prevent clipping.

Pre-amp: Additional gain at user's choice. Works only when Apply gain is enabled (or apply gain and prevent clipping), summed with the gain from the Track (Album) gain tag. Use of this feature is not recommended, because, again, it can negatively affect the quality.
With RG Info- for tracks with ReplayGain tags
Without RG info- for tracks without Replay Gain tags.

Note: if for some reason you do not want or cannot use ReplayGain, there is another option - Advanced Limiter DSP (see below).

2.3.2 DSP

These are digital audio processors needed to perform various transformations of the audio stream in real time.

Ideally, any sound processing should of course be absent, but in some cases to achieve more High Quality you have to use some handlers. In particular, a DSP called Resampler is simply necessary in the absence of hardware support by the sound card for the sampling frequency of the reproduced signal (most often this is material from audio CDs with a sampling frequency of 44.1 kHz) to perform a preliminary conversion of the audio stream into a form corresponding to the hardware capabilities of the sound card (more often in total this conversion is 44.1->48 kHz).

Note: For embedded renderings, the stream is captured immediately after passing through the DSP chain.

Detecting poor oversampling

There is a special sample for this case:

This sample consists of a sequence of tones with a sine wave superimposed on it, the frequency of which floats in the range of 19-20 kHz:

In case of low-quality software resampling, when playing this sample you will hear grinding, noise, or other distortions.

Note: To check resampling accurately, you should first set correct settings output (see clause 2.4.1/2.4.2).

Settings

Let's consider two options for playing materials with a sampling frequency of 44.1 kHz:

a) Your card hardware supports this sampling rate. In this case, to achieve maximum quality, the DSP tab should not have any enabled processors (first list, left).

b) Your card hardware Does NOT support this sampling rate. To achieve maximum quality, install the plugin and set its settings to the maximum sampling frequency that your sound card hardware supports. After the SoX resampler plugin, be sure to add Advanced Limiter to the list of active plugins - this plugin will prevent clipping by “on the fly” lowering the levels of sections of the audio stream that, as a result of signal resampling (or other processing), may be higher than the maximum level.

If you have the second case and you have set the recommended settings, then now it will be possible to play back not only 44.1 kHz recordings with the highest quality (for this sound card), but also recordings with any other sampling frequency. If the frequency of the reproduced material coincides with the maximum supported by your card, the resampler will simply turn off (as unnecessary).

2.3.3 Volume Control

The player's own volume control. If you need software volume control, it is recommended to use it (and not the controls in the sound card/Windows mixer settings). When withdrawing via WASAPI shared ("DS:<звуковая карта>" in Windows Vista/7) is synchronized with the program control in the Windows 7 mixer.

2.3.4 Bit depth converter

Used to convert data into a format that the sound card can understand (usually 16 or 24-bit fixed-point PCM audio). The format selection options depend on the output plugin used (see next paragraph).

2.4 Conclusion

The output plugin is required as a link between the player and Windows/sound card driver. The plugin determines how and through which interface the resulting audio stream (as a result of the work of all previous links) from the player will be output to the sound card. Often this link plays a decisive role, because the use of alternative interfaces allows you to bypass some low-quality sections of the path. It should be noted that at the output stage the stream is most often converted to a fixed-point format (bit depth from 8 to 32 bits - depending on the selected parameters), what this can lead to has already been discussed in the section on ReplayGain. On this moment foobar2000 has the following audio output methods: DirectSound, Kernel Streaming, ASIO, WASAPI Shared, WASAPI Exclusive (only for Vista and Windows 7).

Due to the fundamental differences between the architectures of Windows XP and Windows 7, we will consider setting the output for each OS separately.

2.4.1 Windows XP

Here is an approximate diagram of the Windows XP sound subsystem:

As you can see, when outputting through DirectSound or MME, the sound goes through the Windows mixer (Kmixer). The main difference between DirectSound is the wide range of possibilities for using the hardware resources of the sound device, incl. hardware mixing and low latency. But the fact is that these capabilities directly depend on the drivers used. Thus, due to low-quality sound card drivers (this is not uncommon now, especially outdated drivers) audio may be distorted. They can arise either as a result of the operation of the Wave regulator (included in Kmixer), or as a result of poor-quality software resampling or other unwanted processing.

Two other outputs that can bypass Kmixer will help solve the problem - ASIO and Kernel Straming. Kernel Streaming is a bit-by-bit audio output tool from Microsoft, part of DirectSound (in the diagram - a stream bypassing Kmixer). ASIO also allows you to bypass Kmixer, but first of all it is a professional input/output standard designed to minimize latency, although this is not important for high-quality music playback.

Different ASIO drivers may contain their own errors, and therefore it is recommended to use Kernel Streaming to achieve the most accurate output.

Setting:

And so, everything is simple here. Install the plugin, run foobar2000 and on the output page select:

Output Device:KS:<Ваша звуковая карта> - the Kernel Streaming plugin will be used to output sound, the sound will be output to the specified sound card.

Buffer Length: adjusting the player buffer. How less value- the faster (counting from the moment of inclusion/change) volume changes, inclusion of plug-ins, etc. will take effect. This option does not affect the output quality; changing the standard value is not recommended.

Output format/Postprocessing

Output data format: set the maximum bit depth supported by your sound card.

Dither: enable dither + noise shaping. This function can only be useful if one of the following conditions is met:

a) If your sound card hardware does NOT support audio bit depth higher than 16 bits and you use one of the functions: ReplayGain, DSP, Volume Control (volume control in foobar2000).

b) Your sound card hardware does NOT support audio bit depth higher than 16 bits and you are playing material with bit depth higher than 16 bits in foobar2000.

In all other cases (for example, if your sound card hardware supports audio bit depth higher than 16 bits) Enabling this feature will only degrade the quality.

2.4.2 Windows 7

In this OS (as in Vista) sound subsystem has a completely different structure. Below is a simplified block diagram of it:

API- Application Programming Interface
APO- Audio Processing Object
CPT- Cross Process Transport
KST- Kernel Streaming Transport

As you can see from the diagram, there is no DirectSound here (there is only its visibility for compatibility with older programs). By default, all sounds are output via the WASAPI interface ( Windows Audio Session API) in the so-called general mode, which includes various services, software processors and a mixer. It is also obvious that the sound is transmitted to the device driver only after passing through all the above components. Thus, all sounds are reduced to the same frequency and bit depth, mixed (all processing is carried out using the computing resources of the CPU), and the stream arrives to the sound card in finished form.

As practice has shown, such processing narrows the dynamic range of a 16-bit signal by about 2 dB, which is equivalent to introducing a little noise into it. But we don't need noise, do we?

Note: A caveat needs to be made here. If the settings of the output device (in the control panel) set the bit depth to 24 (of course, the device must support such a quantization depth), the level of introduced noise is minimized. But nevertheless, we still won’t be able to get bit-by-bit output.

This OS still has Kernel Streaming (KST), but in practice it does not always work - sometimes the device is “busy” for unknown reasons.

If you look at the diagram, you will notice that direct access to the driver is available via ASIO. Hence the conclusion: if your card supports ASIO in hardware (and has a sane ASIO 2.0 driver) - use this interface. Don't forget that this requires a plugin.

But what to do if the card does not have ASIO support? I’ll say right away that ASIO4ALL is unlikely to help here, since it works through the same unstable Kernel Streaming.

Fortunately, there is still a way out of this unpleasant situation. The developers left us a loophole, and it’s called WASAPI Exclusive Mode. In this mode, you can gain direct access to the hardware resources of the sound card and bypass all unwanted WASAPI components. However, in exclusive mode there are strict restrictions - when an application uses this mode, the sounds of all other applications are turned off.

Setting:

And so, download the WASAPI output plugin, install and configure the output as in the screenshot. In field Output data format you need to set the maximum bit depth at which playback will work (most often - 24 bits). About the function Dither read in the previous paragraph (Windows XP).

3. Setting up the sound card

Since there are a huge number of sound cards, this part provides only general recommendations or recommendations for setting them up using standard means Windows.

3.1 Setting up using Windows

3.1.1 Windows XP

To start setting up, you need to go to Windows Control Panel->(Sound, Speech and Audio Devices->)->Sounds and Audio Devices.

Rice. 1.1 The volume should be set to maximum, the volume level should be adjusted using the analogue method - for example, using a knob on the amplifier (if you use headphones without a volume control, adjust it using foobar2000). After completing the settings, press the button Additionally in the column Mixer volume.

Rice. 1.2. All sources except Wave (which must be set to maximum) and Play Control, of course, should be turned off. Wave can also be disabled if you do not use DirectSound output in the player, but all other sounds, including system sounds, will not be heard. First you need to go to "Properties" and enable the display of all sources:

Rice. 1.3. Here you must enable the display of all playback sources for the device being configured.

Rice. 1.5. Here both regulators should be set to the maximum position. Naturally, it is assumed that you do not experience any problems with sound playback (most often this can happen in games). If there are problems, select the highest positions of the regulators at which the problems will disappear. In this case, it is worth noting that these settings do not affect playback using Kernel Streaming and ASIO, it follows that if you had to set one of the controls to a lower position, for quality output For sound in the player, you must use one of these output plugins.

After completing the settings, click the button in each window OK.

3.1.2 Windows 7

To start setting up, you need to go to Windows Control Panel->Hardware and Sound->Sound and on the first tab, double-click on the playback device that you want to configure.

Rice. 2.1 Set the main volume to maximum. We turn off all unnecessary sources (microphone, line input), set the channel controls (Front, Rear, etc. - they are not visible in the screenshot) to the maximum position..

Rice. 2.2 The presence and name of this tab depends on the installed sound card. As experiments have shown, enabling/disabling this option only affects software effects that are superimposed when outputting in general mode. When outputting via WASAPI Exclusive, this option does not affect anything - in this case, only the effects imposed by the hardware DSP of the sound card will work.

Rice. 2.3 On this tab, we are most interested in enabling the exclusive mode required for WASAPI to work. Purpose of the option Give priority to exclusive mode applications I don't know for sure, but it's better to leave it on.

Setting the frequency and bit depth here also works only for the general mode, i.e. This is the reference frequency and bit depth of the software mixer to which all sounds entering it will be reduced. It is advisable to set the frequency corresponding to the characteristics of your sound card, the bit depth - the maximum supported (use the Examination).

At this stage, the setup principle is no different from the previous ones: it is necessary to ensure reliable transmission of the digital audio stream without making any changes to it (ideally, up to the digital-to-analog converter itself). Setup recommendations are as follows:

• All effects/equalizers, etc. must be turned off. If possible, you should enable the mode Bit-Matched Playback(for Creative cards, this option may be in the Creative Audio Console or Console Launcher).


• For most cards, it is possible to select a reference sampling rate (Master Sampling Rate) - it should be set equal to the sampling rate of the reproduced material or the frequency set in the foobar2000 resampler. There may also be a setting for the quantization depth (Bit Depth) - it should be set to maximum.


• All volume controls that affect playback volume should be set to maximum.

For getting best result, it is recommended that you do the following before playing a track.

For lossy:

1. Select the track(s) in the foobar2000 library or playlist, right-click on the selected one and select from the drop-down list Utils->Verify Integrity. The track(s) will be scanned for errors. If there are errors in MP3 files, foobar2000 can fix them. If you see an error like “Reported length is inaccurate...”, select the tracks for which the error was displayed this error, right-click and select Utils->Fix VBR MP3 Header, after processing, check the track again, there should be no error.

If, when checking, you see an error like “MPEG Stream error...”, you can fix it using the Utils->Rebuild MP3 Stream option. After performing the error correction operation, do not forget to check the tracks again.

2. Select tracks, click on the selection right click and select ReplayGain->Scan Per-File Track Gain if tracks from different albums are selected or albums are not specified in tags, Scan Selection As Single Album - if tracks of one album are selected and Scan As Albums (by tags) - if several albums are selected. You can also use the Scan As Albums (by folders) option if album tags are not specified but the tracks of each album are in a separate folder. After scanning you will see information about the calculated peaks and gain values ​​for each album/track (depending on the selected scan type), click the "Update File Tags" button to write the ReplayGain tags to a file.

For lossless:

With this source type, no action is required to achieve maximum quality. You can also check the material for errors using the Utils->Verify Integrity function, but the probability of errors for lossless is less, especially if there are any, there is no way to fix them using foobar2000 (at least in the same way as in the case of MP3 ). There is no need to use ReplayGain in the case of lossless, because clipping (in the form in which we usually have it in the case of lossy) cannot exist in lossless sources, so this function can only be useful for leveling the volume between tracks/albums.