Peak normalization of audio signals resulted in significant loudness differences between broadcast channels;

The indications of the QPPM level meter, standardized in European countries by EBU Tech Doc 3205-E and commonly used, do not reflect the loudness of the signal, because this device was not originally intended to record the average value of the signal;

With the rapid growth of digital production of phonograms and digital distribution of audio content, the rationing of the permitted maximum audio signal level, defined by the document ITU-R BS. 645, does not meet modern requirements and has become obsolete;

Document ITU-R BS. 1770 defines an international standard for measuring the loudness of audio programs, introducing new parameter audio signal - a unit of loudness.

In accordance with the above, the European Broadcasting Union recommends using the new level unit LU (Loudness Unit) and LUFS (loudness relative to full scale) when measuring audio signals. (The name “LUFS” corresponds to the international terminology convention and is equivalent to the name LKFS, which is used by ITU-R BS.1770-2).
Recommended for complete characteristics transmission to make measurements on three main parameters:

- Program Loudness;
- Loudness Range;
- The maximum instantaneous level (Maximum True Peak Level).

The basic rules for measuring these parameters are as follows:

EBU R 128 recommends taking a level equal to -23 LUFS as the nominal value of the program loudness, and in cases where the exact maintenance of the nominal level is unattainable (for example, during a live broadcast), the permissible deviation from the nominal level should not exceed ± 1.0 LU.

The audio signal of a transmission should generally be measured as a whole without isolating specific fragments such as speech, music or sound effects.

The maximum allowable instantaneous transmit level should be -1 dBTP (Decibels True Peak).

All measurements must be made with meters specified in the relevant documents: ITU-R BS.1770, EBU Tech Doc 3341 and EBU Tech Doc 3342.

*EBU - European Broadcasting Union (European Broadcasting Union)

For reference, only Channel One, VGTRK, Radio Mayak, Orpheus, and Voice of Russia are members of the EBU (EBC) in Russia. What standards other broadcasters use is anyone's guess.

Attached is an archive with EBU documents in Russian, namely:

EBU Tech 3341;
EBU Tech 3342;
EBU Tech 3343;
EBU Tech 3344;
Essay_625v2- essay by Anatoly Sokolin: "The revolution that shook the world of audio";
R68_2000_EBU- technical recommendation EBU R68-2000. Setting level in digital audio production equipment and digital audio recorders;
EBU R1771- requirements for instruments measuring loudness and true peak level;
EBU R1770-1- Recommendation ITU-R BS.1770-1. Loudness measurement algorithms sound programs and true peak audio signal level;

Here you can always get up-to-date original documents.

Music lovers, we believe, have often had to deal with a situation where some compositions of one collection sound too loud, while others, on the contrary, are too quiet. This happens when users download music from different sources, and even compositions with different sound volumes are often found in mixes that contain songs from different artists. Well, this is understandable, but how to act in such cases, not to adjust the volume every time, as soon as the song sounds louder or quieter?

No, of course not, because the volume can be equalized, and very simply. For this you need a small free utility. This program allows you to process audio files of popular formats in batch mode in accordance with set parameter volume.

So, go to the developer page and download the latest full (!) version . By default, the program is installed on English language and if that doesn't bother you, set it to normal mode in order to immediately get the interface in Russian, at the second step of the installation, you must check the box Russian on the menu "Language files".

Press the button in the menu "Add Files" and load into the utility window audio files, the volume of which needs to be worked on. Next click "Analysis track" and wait until the program finishes analyzing the files in order to identify their volume. This procedure may take several minutes, it all depends on the total size of the files being analyzed. It takes about 10 minutes to process a 1 GB assembly.

After you need to set the desired volume (default is 89 Db) and press the button "Track type". As a result of processing, the volume of all audio files will be reduced to a single specified value. Changing the volume level takes less time than analysis. It is also worth noting that all files in the process of processing and saving will be overwritten.

And finally, a few words about what the parameters in front of each file mean.

• Level- current volume.
• clipping- opposite bird Y indicates that there is noise in the background of the track at the current volume (whether you hear them or not depends on your hearing acuity).
• Track- shows the difference in decibels between the current and the user-defined volume setting.
• The presence of a mark in the column "Clip (T)" indicates that background noise will remain after processing.

To keep these noises as low as possible, it is not recommended to set the volume too high or too low. The optimal value for the parameter "Norm of loudness" is approximately 85-95 decibel.

Sound Normalization in Sound Forge Pro 10

Raise the level of the audio signal without the risk of losing quality, allows the function "Normalize". The algorithm of its work is as follows: the program subtracts the level of the highest signal from the level of the maximum possible signal, raising the overall volume level of the file by the resulting difference. To take advantage "Normalize" function Let's open the dialog box of the same name, located in the "Process" menu item. The main parameter is Normalize to, indicating the maximum possible signal level that will be taken into account when sound normalization in Sound Forge.

It is possible to normalize the signals of several files, which is useful when burning a CD. To do this, by pressing the button "Scan levels", we will scan an audio file, the volume of which will be equal to the rest. Then open the following audio file and in the "Normalize" dialog box, check the box next to the radio button "Use current scan level (do not scan selection)". Click the "OK" button in the dialog box Normalize to. The program will produce volume normalization in an audio file.

Function "Normalize" can also perform more complex processing by calculating the average "perceived loudness". Sometimes a situation arises when a sound in an audio file, being at the same volume level as the rest, sounds louder. The reason for this is the properties of human hearing. Sound Forge has the ability to measure file data in terms of human perception. To do this, in the "Normalize to" dialog box, you need to activate the switch . In this case, several more options will be available: "Ignore below"- the value of this parameter determines the threshold of the acceptable sound level. All values ​​below the specified threshold will be ignored when scanning "Average Perceived Loudness". In most cases, the value of this parameter is approximately "-45 Db".

Parameter Attack time tells the software how fast to open the digital signal gateway to allow for acceptable audio levels when scanning data. Therefore, if the audio file contains frequently changing sounds, such as drumsticks, you should set this value as low as possible, otherwise frequent sounds will not be taken into account. In most cases, a value of 200 milliseconds will do the job.

Parameter Release time tells the program how fast to close the digital gateway. If you want to take into account as much material as possible when scanning data, you should set this parameter to a higher value.

Due to some limitation of human hearing, very high and very low frequencies harder to hear than average. This situation can be corrected by checking the box "Use equal loudness contour". This function enhances inaudible spectra in frequencies, so in most cases it is advisable to set it.

After setting all the parameters, press the button "Scan levels" to start the scanning process "perceived loudness".

When working with the function "Average RMS level (loudness)", you should be careful when setting "Normalize to" as selecting a very high value may result in sound distortion or data clipping. If you do not exceed the value "-6 Db", distortion is excluded.

For better protection against data pruning, you can select "Apply dynamic compression" bookmarked "if clipping occurs".

Press the "OK" button. The program normalizes the volume of audio data, taking into account the values ​​of the current parameters.

The computer program allows you to equalize the volume of MP3 music files. The first version of this utility has existed since 2002. The program is good because it does not require transcoding files at all - this allows you to maintain the original sound quality. MP3Gain equalizes the volume level of both a single file and an entire group of files (batch conversion).
We will not dive into all the subtleties of the settings and capabilities of the program - we will just learn how to easily normalize the volume level in mp3 files without unnecessary problems.
We find and .

All the advantages of the program
The program is completely free.
Installed on any version of Windows OS.
Can be used and operated command line and graphical shell for Windows.
Possibility of batch analysis and file processing.
Normalization occurs without recoding files.
You can convert the same mp3 file many times without the risk of damaging it.
There is a mode of applying normalization only to the tracks selected in the working window.
The program fully saves ID3 tags and creation dates of files.
Multilingual interface, including Russian localization.
Localized reference guide on the official site.

Installing MP3Gain
We take the program from SourceForge in the form of an installer. Installation is extremely simple, the only important point- it is necessary to enable the checkbox "Language Files", while all language localizations of the program will be installed, including Russian. If you choose "Custom" installation, you can choose the program's parking directory yourself.

Setting MP3 Gain
After installing the program, run it and first of all select the Russian localization of MP3Gain. Next, open the experimental mp3 files. In the program settings, we are looking for a very important item “Changing the level without clipping” and put a checkmark on it. For short, "clipping" is the excess of the signal level, while cutting the level and re-encoding mp3 files, but we do not need this. And still it is necessary to disassemble the question of setting the volume level. By default, the "Normal" volume is set to 89 decibels (it is better not to change this figure). According to experts, 89.0 dB gives the highest quality results in terms of normalization and elimination of clipping. The rest of the settings are very clear and set according to personal preferences or just do everything as shown in the picture. These settings are enough for simple normalization of the volume level in mp3 files.

Advice! Just in case, you need to make copies of the audio files. MP3Gain does not have a function to save processed files under a different name, the program overwrites the original ones.

Using MP3Gain
To understand what to do with the two working buttons "Analysis" and "Type", you need to briefly understand their available modes.
Consider the modes "Track", "Album" and "Constant".
Track- the program calculates the volume level, individually for each track. It then adjusts the volume of each track to the desired level. For example, there are three songs with a volume level of 87, 95 and 91 dB. If you use "Track Type" to bring them to the desired level of 89 dB, all of these songs will output about 89 dB.
Album- The overall volume of the album will be adjusted to the desired level, but the difference in volume between the tracks in the album will be preserved. For example, if there are three songs with volume levels of 87, 91 and 89 dB, the total volume of this album will be about 89 dB. By applying "Album Type" to bring them to the required level of 92 dB, the program will increase the volume of each of these songs by 3 dB.
Constant- This mode is similar to Album mode. With it, the volume of all tracks is simply increased or decreased by a given number of decibels without any normalization relative to each other.

So, let's conduct an experiment on previously opened mp3 files in the "Track" mode. First of all, we start the analysis of files with the "Analysis Track" button. We look at the result of the analysis source files. The picture below shows that in the files "3.mp3" and "5.mp3" there is an excess of the volume level, in other words, there is "clipping", the letter "Y" appeared in these columns and all lines turned red. On the contrary, in the file "6.mp3" you can see that it has a reduced volume level.
And then, based on the results of the analysis, the second step is to correct (normalize) this level difference by pressing the "Track Type" button. After successful normalization, which took some time (it all depends on the power of the computer), we look at the resulting result. The last picture shows that the level of all processed mp3 files is very close to the target value of 89 dB. Those. the tracks "3.mp3" and "5.mp3" lowered their volume level, while the track "6.mp3" increased it. Which is what needed to be done!

So it is with sound files, only they do not hear, but store. Some files store sound with a volume level conforming to the norm. Others store sound at a volume level deviant. However, this does not change the volume level of the original audio signal. Everything is defined only recording level sound signal. And the recording level, in order to prevent distortion, is set in such a way that sound signal, supplied to the input of the analog-to-digital converter (ADC) was slightly below its maximum possible level at maximum volume values ​​(peaks). Otherwise distortion cannot be avoided and the sound becomes unnaturally distorted. In addition, the recording level can be lower even in a variety of technical reasons.

For a person norm hearing sensitivity is determined range or two positions corresponding to the lower and upper sound levels - this is the so-called dynamic range or hearing area. One position corresponds to the weakest sound volume level (sensitivity threshold) still distinguishable by a person. The other position corresponds to the strongest level of sound volume (pain threshold) still perceived by a person.

There is also a norm for sound files, but this volume recording level. It is determined range or two positions corresponding to the lower and upper levels of the sound volume recording and this is also the so-called dynamic range. Since the computer understands and processes sound in digital form, it stores it in files also in digital form. And so the range is defined by two numbers corresponding to the upper and lower bounds. dynamic range. Depending on the stored sound quality the range will have a different width.

Speaking further about sound files, we will mean files with the extension .WAV, that is wav files. Since these are the files that are used for recording audio CDs.

On the audio CDs sound files are stored with the extension .RAW. When written to a hard disk, they are converted into wav files. When burned to an audio CD, audio files with the extension .WAV converted to files with the extension .RAW.

Quite often, the recording volume level of the sound at audio CDs for different musical compositions is unequal or lower norms, which creates an unpleasant sensation when moving from one musical composition to another. It also often turns out unequal and values ​​for the recording level of sound volume for two stereo channels of one musical composition. To overcome these shortcomings, it is necessary to normalization. Which was invented for this.

If not fulfilled normalization, then the low volume level of the sound recording of a musical composition when playing an audio CD will require compensation in the form more gain from the audio equipment. Which is quite inconvenient and leads to the emergence of completely not necessarily misrepresentations from the audio equipment. which could be avoided by using non-distorting sound processing in the form of normalization.

As a possible tool for normalizing wav files, I would like to recommend you a shareware program. This program allows you to process wav files with the usual 8 and 16 bit digital format and one or two stereo channels. By opening and processing wav sound files, Sound Normalizer 2.2 creates copy original sound file, with which it works. Thus allowing in critical situations to avoid irreversible change original file. Also one of the advantages of the program is the possibility independent adjust the sound normalization level for each channel. Unlike other similar programs, Sound Normalizer 2.2 has simple and clear interface.

Normalization is carried out according to peak or maximum sound levels. And this means that each value of the sound level will be subjected to proportional change and thus keep natural sound the entire musical composition.

However, it also happens that the sound for two stereo channels, after normalizing to the maximum level, will be perceived as having various volume levels. This is because the perception of loudness has more dynamic and frequency components that can be so different between the two channels as to cause the average volume level for each channel to be heavily skewed. In this case, it can be recommended to perform the trial and error method manual adjustment normalization level for each channel.