Well, if you tell everything and about everything, then you need to talk for a long time and a lot, so I’ll only tell you about some of the most interesting functions for cleaning recordings from noise.
Let's assume that we already have an audio file open, from which we need to remove tape hiss. To do this, we select a fragment of the recording in which only one tape hiss sounds, without a useful signal. This could be a pause between songs, or pauses at the beginning and end of a recording. Therefore, we should not rush to remove pauses; they will help us a lot when clearing the recording of noise. For a more accurate and accurate selection, stretch the wave image horizontally so that you can see the selected fragment on the entire screen and do everything carefully, without capturing anything unnecessary in the selection. Unfortunately, I haven’t figured out how to do this using JAWS yet, if anyone can explain, I’d be very grateful! To change the horizontal scale of the wave, you can use the mouse wheel or special zoom buttons in the Adobe Audition window.
Now open the “Effects” menu, select the “Noise reduction” item and in the submenu that appears again the “Noise reduction” item. The noise removal window will open. In new versions of Adobe Audition, this option opens through some other menu, but in version 1.0, which I use, the noise removal function opens this way.
A few words about the settings, the values of which I would recommend changing and setting them to other values that differ from the default settings.
First is the "FFT" combo box. If the value in this list is very small, the useful signal will suffer significantly when removing noise. If the value is very large, recording processing will take a long time, the useful signal will not be affected, but the cleaning efficiency will be lower. I recommend setting the FFT value to 6000, because as a result of my experiments with sound cleaning, I found out that for clearing the most common noise, such as tape hiss or noise from a not very high-quality sound card, this is the best option.
Secondly, “Accuracy”, by default it is set to 7, for stereo recording it is better to set the value to 14 or 15, this is already quite a lot, in my opinion, there is no point in setting it more, since apart from increasing the processing time, it does not give anything . And thirdly, “Spectrum Attenuation”. In this setting, the default value is, in my opinion, 56, I recommend setting it to 0, since the higher this value, the less effective the noise removal. An exception can be made in cases where the tape hiss is very strong in comparison with the useful signal. Then, in the attenuation of the spectrum, you can try to increase the value in order to get rid of some kind of “metallic” chime that remains after removing such strong noise.
So, the noise that should be removed from the recording is selected and the noise removal window is open. Click the “Get from selection” button in this window and wait for a while. A scale appears on the screen showing the percentage of the process completed. When the scale reached one hundred percent and disappeared, everything was ready, we provided the program with a sample of noise that needed to be removed from the entire recording.
There are two buttons in the noise removal window, “Ok” and “Close”. Click “Close”, since pressing the “Ok” button will start the cleaning process, and we don’t need this yet. You should also not close the noise clearing window by pressing Alt+F4, clicking on the cross or clicking the "Cancel" button, as this will lead to the cancellation of everything we did in the noise clearing window. After the window is closed, we need to select the fragment of the recording to which noise cleaning will be applied.
Here I would like to draw your attention to the fact that all Adobe audition functions apply only to the selected fragment of the recording. If a fragment is not selected, any launched operation will be applied only to that part of the soundtrack that is currently on the screen. Therefore, if we want to apply the noise removal function to the entire recording, after closing the noise removal window, press ctrl+a to select the entire audio track. After this, using the method already described above, open the noise removal window, press the “test” button and listen, our recording will sound the way it will sound after processing. If we are satisfied with everything, click the “Ok” button and wait for the processing process to complete.
Another important point: it is possible that after the first cleaning the noise will not be completely removed. The noise removal function has one property: the further we go in the recording from the place where we took the noise sample, the more this same noise that we removed begins to appear. I don't know what this is connected with. Either the noise changes slightly over time and no longer coincides with the resulting sample, or with the noise cleaning function, the further you go, the greater the mismatch with the noise. Therefore, it is possible that in order to completely get rid of noise, after processing you will have to once again select a sample of one noise, in the place where it remained, and repeat the entire operation. The noise removal function works very well on homogeneous or cyclically repeating noise. To remove, for example, the crack of a record, I would highly recommend first using the Clip pop eliminator function, and only then removing what remains using the noise removal function.
Especially for removing hiss, Adobe audition has a hiss removal function, but it seems to be based on the use of some kind of frequency filters; when using it, the hiss is not particularly removed, and the quality of the main signal suffers. Therefore, if there is a place on the phonogram from which you can get a sample of one noise without a useful signal, then to remove hiss it is better to use the noise removal function by sample rather than the hiss removal function. If you have made a recording, for example, from a cassette or gramophone record, I strongly advise you to cut one song from it and process each composition separately, and not the entire large recording at once. This way you will achieve better results.
After noise removal is completed, you can remove all unnecessary pauses at the beginning and end of the recording, apply some other functions if necessary, for example, sometimes it is very useful to correct the frequency characteristics of the recording using an equalizer. And as the final stage of recording processing, you can perform normalization.
Normalization
A little about what types of normalization there are and what it is all about.
Normalization is bringing the recording volume level to a certain value, for example, to the maximum permissible volume level. After normalization is completed, firstly, your recording will not sound quiet; if you wish, you can turn it on on any computer or player and use all the capabilities of its acoustics in terms of volume. And secondly, if you recorded several tracks and normalized them all to the same volume level, they will all sound the same and it will not turn out that one song sounds louder and another quieter.
The first normalization option is peak normalization. The principle of its operation is as follows. The program analyzes the entire recording and finds those places in it where the volume reaches the maximum value, and after that, increases the volume of the entire recording so that these places reach the maximum permissible value.
Peak normalization works very well for classical music. But it also has a minus. For example, in our musical phonogram there is, say, a drum, which at some point banged so hard that it sounded much louder than all other instruments. Naturally, this sound in the recording is the loudest, peak normalization will focus on it and increase the volume of the recording so that this fucking drum reaches the maximum allowable volume value. But all the other instruments sound much quieter, as a result, the recording turned out to be quiet, only this very drum banged loudly enough once. Yes, such normalization will not introduce any distortion into the sound, but if you come across such a drummer, you will not be able to get the desired recording volume, but, for example, as I already wrote, for classical music this is a good option, since it will not introduce any other changes to the sound except changing its volume.
In Adobe audition, you can use peak normalization in the following way: select the entire recording by pressing ctrl+a, open the effects menu, select the “Amplitude” item in it, and “Amplify” in the submenu of this item. In the window that appears, on the “Constant amplification” tab, there are one or two volume controls, depending on what kind of recording we have, mono or stereo. You can move the controls manually, or you can press the "Calculate" button. Then the program will analyze the recording and itself set the controls to the value, in this case necessary for peak normalization of the recording.
After this, all you have to do is click the “ok” button. It’s also worth paying attention to the “Left right block” checkbox. If the checkbox is checked, when calculating the peak normalization level, the program will focus on the loudest sound in one of the two channels, but will set the volume controls so that it does not change the volume ratio of the left and right channels. If the checkbox is unchecked, each channel will be normalized separately.
The second type of normalization is. Um, I'm sorry, it's a bit confusing with terms, I forgot what it's called. But its essence is that it does not just change the volume, but, apparently, processes the numerical values of our digital recording in some cunning way. The gist is something like this. For example, a balalaika and a violin play simultaneously, but the balalaika player plays so loudly that he completely drowns out the violin. After our second type of normalization with a forgotten name, :-) both the balalaika and our muffled violin become more or less audible. Now, our “drummer” can drum as loud as he wants, he will no longer be able to drown out everyone else.
Personally, I think this method is suitable for most cases and is very good for processing digitized concert recordings. For example, if someone shouted loudly into the microphone, this scream, after normalization, will no longer seem too harsh and significantly louder than all the others. At the same time, such normalization does not mean at all that all noise will be blown out, especially during pauses. Not at all, this type of normalization does not change the volume level in different places in the recording, the recording gain is the same throughout the entire recording.
Although, if there is noise in the recording that was previously drowned out by the useful signal and was not heard, it may take a more obvious place against the background volume of all instruments.
In Adobe audition, this normalization can be used by selecting the “Amplitude” item in the “Effects” menu, and the “Normalize” item in the submenu of this item.
The third type of normalization is the so-called strict limitation. It operates on the principle of AGC (or ARUZ), radio amateurs will understand me. This type of normalization changes the gain level of a recording depending on the current loudness of the original recording. For example, if someone barks loudly into the microphone, the program will lower the volume at that moment, and when someone speaks quietly, the program will increase the volume so that they can be heard better. Using this for music, of course, is a complete outrage on the phonogram, but in cases where several people are speaking, one sits far from the microphone, another is closer, and the third is very close, this method of normalization can help, but only under the condition that there is no noise on the recording and the interlocutors are talking in silence, otherwise this will also be an outrage, since in the pauses between words the noise will mercilessly come to the fore (you can set the input signal level, starting from which this type of amplitude correction will operate, and thus eliminate noise from the process, but then the quiet voice of one of the interlocutors may “fall out.”
In Adobe Audition, strict amplitude limitation can be applied by opening the “Effects” menu, selecting the “Amplitude” item in it, and the “Hard limiting” item in its submenu. The settings for this function are made by analogy with a hardware device. "Input signal level", "Search speed" ("rise rate") and "decay rate" ("fall rate").
After everything is cleared of noise, edited, processed with other effects and normalized, you can save the recording. The only thing is that noise removal, especially the click removal function, must be applied before normalization.
Well, that seems to be all the basics about cleaning sound and bringing it to normal sound in Adobe Audition.
In the understanding of uninitiated users, audio editing means converting discs to MP3 or FLAC, as well as cutting tracks from a disc image with a CUE markup file. The most advanced “pros” can even make ringtones by cutting out a fragment they like from a musical composition. These are, of course, quite popular actions, but the term “sound processing” has a completely different meaning.
How can I change the sound? First of all, you can influence the frequency characteristics of the audio signal. Frequency correction is akin to a sharpening filter when processing an image - visually the object is made clearer and more legible, but this does not mean that details have been added after editing. When processing audio signal frequencies, approximately the same thing happens. I suppose everyone who wore headphones with a player sometimes felt the need to “tweak the equalizer” to add bass or pull out the high frequencies. But what’s interesting is that strongly protruded high frequencies increased noise, which did not decorate the audio track. The reverse procedure—noise suppression using frequency equalization—has a different problem. Along with the noise, a certain spectrum of vocals and music disappears, which leads to the loss of useful information.
Another way to “change” the sound is through the use of effects. In our opinion, such a sound processing technique is justified only when the sound engineer is working on mixing audio material and has a clear idea of what he is doing. Sometimes even a distortion filter can make an audio fragment sound natural. In addition, audio filters often help to hide or veil some defects in the audio track. For example, if there is a strange sound or echo at the end of a recorded instrument part, you can correct this by cutting out or muting the problem area and “immersing” the ending in a smooth echo with a decay effect.
There is a third way to edit digital audio. This is what the sound processing principle in most modern audio editors is based on. For the original recording, a waveform is created - a diagram from which you can visually trace the change in sound amplitude over time. If you open any audio file in a sound editor, such as Sound Forge or Goldwave, you will see that this is the default sound representation option in the program. Modern sound editing packages allow you to make changes to the shape of this “sine wave” - change the shape of the peaks, compress or expand them, remove fragments and modify the nature of the curve. This opens up wide possibilities for the user who sets out to process sound: in the original audio material in the wave form, you can easily find clicks and noise, and then remove them, etc.
Noise is one of the most pressing problems encountered in audio processing. Noise is a natural phenomenon. It is present everywhere and in everything - water gurgles from the tap in the kitchen, cars make noise on the street, computers and printers hum in the office. Accompanies noise and any audio recording, be it a message on an answering machine or musical compositions from an audio disc. Of course, the noise level in all these cases varies, however, it is impossible to completely get rid of it. The nature of noise can be very different - from the buzzing of a drill to the rustling of leaves, so it is impossible to come up with an ideal algorithm that would clear sound from noise. In addition, audio interference is akin to dust and scratches in a photograph - artifacts can be “masked” as the main sound and made less noticeable, or rather inaudible, and can be removed completely.
The most important question that faces users is which tool to choose to combat noise? The choice of a sound processing tool depends on financial preferences. Of course, there are professional applications that use unique audio processing algorithms and cost thousands of dollars. The capabilities of specialized software can be much wider and not limited to noise removal alone. For example, some apps may remove certain sounds from the audio track, such as the occasional noise of a passing car. However, this option can only be afforded by large companies, say, film studios, which have a fairly substantial budget at their disposal. For a user with an average income, a pretty good option is Sound Forge Pro or Adobe Audition. Well, as an economical option, we can offer it, but it must be the latest version, even if it is a beta version. Don't be afraid of the editor's test version. In fact, this program works very stably and is no worse than the official release. In addition, it has some tools that were clearly missing in the previous version.
⇡ Removing noise in pauses between audio tracks in Sound Forge Pro
This professional audio processing package contains two key tools to help you beat noise. The first one is called Noise Gate. Imagine an old recording from an audio cassette, where a sharp hissing sound is heard during pauses. Perhaps someone will consider this noise an indispensable attribute of the music of the eighties and will not want to get rid of it when digitizing rare recordings. But if you don’t find any romance in this noise, you can easily suppress it, and without harm to the main composition. To do this, a filter is used that removes the signal if it drops to a certain threshold value.
Open the audio file in Sound Forge Pro and select the section of the track that you want to process. Now select Effects and Noise Gate from the program menu. Like most other tools in the application, the filter has several presets and the ability to preview the result with the selected settings.
⇡ Removing background noise in Sound Forge Pro
Sound Forge Pro's second noise reduction tool is the Noise Reduction module, which can be accessed from the Tools menu. The presence of noise usually means that the original information has been lost. The computer, when eliminating random interference, must somehow recover this data. However, by and large, this is impossible, since removing frequencies containing noise leads to the loss of the original sound and the appearance of gurgling artifacts.
When noise is random in nature, it is difficult to describe mathematically. It’s another matter if this noise is caused by cyclic processes, say, electromagnetic oscillations. In this case, you can predict how the waveform should be adjusted to obtain an undistorted signal at the output. To combat these "predictable" types of interference, a background noise capture technique is used. Its essence boils down to the fact that the program records a fragment of “pure” noise and then subtracts it from the audio recording. Such a fragment can be very short, lasting less than a second.
Sony Noise Reduction allows you to either automatically capture noise or do it manually. In manual suppression mode, the module generates a picture of the frequency spectrum and marks control points on it that determine the nature of noise suppression. Those data on the graph that are located above the envelope will be left unchanged. And to those located below it, attenuation will be applied. Accordingly, when manually moving control points up, the noise suppression effect will be stronger, and vice versa.
Sony Noise Reduction can use one of four noise reduction algorithms, which are called Mode 0, Mode 1, Mode 2 and Mode 3.
The default mode is Mode 2, which works well for most situations. But the developers recommend that you always check how the program copes with noise removal in all possible modes and choose the best result.
- Mode 0 is generally suitable for eliminating low amplitude noise. However, its use increases the likelihood of distorting artifacts;
- Mode 1 is similar to Mode 0, but is less likely to cause artifacts. On the other hand, it removes less noise;
- Mode 3 is the most gentle noise reduction mode, in which the likelihood of sound defects is minimized.
⇡ Removing clicks from audio recordings in Sound Forge Pro
Typically, a music recording transferred to digital form from a vinyl record is accompanied by crackling and clicking sounds. In this case, it is not enough to use a noise profile because the record crack is unpredictable. Sound Forge Pro has a special tool for eliminating the most common noise when digitizing audio recordings. To call it, run the command “Tools → Audio Restoration”.
In addition to selecting the noise threshold and other background noise suppression options, this tool also gives you the ability to adjust click suppression. The Click Removal Amount slider is responsible for this. By default, its value is set in the middle and is optimal for most records. Moving the slider to the right will remove more clicks, but it may also accidentally remove elements of the main audio, such as a drum sound.
⇡ Removing background noise in Audacity
To remove noise in the free Audacity editor, you can use the same techniques as in the case of Sound Forge. For example, the “Noise Removal” module, written by Dominic Mazzoni, in fact, one of the authors of Audacity, is suitable. To use this filter, you need to take a few seconds to create a noise profile, and then choose Effects → Noise Removal.
After this, you need to click on the “Create noise model” button and, having selected the entire audio fragment to be processed, call this filter again. Now in the settings window you can select the strength of noise reduction (Noise reduction), adjust the filter sensitivity, specify the smoothing frequency and attack time, which characterizes the filter’s reaction to the appearance of noise.
Sometimes the noise profile method does not work, or the recording is so dense that there is nowhere to get this profile from. In this case, you can resort to this trick. To obtain a profile, you simply need to make an additional recording under the same conditions in which the edited material was recorded. Next, the profile can be taken from a new entry. Unfortunately, it is not always possible to recreate the initial conditions, so this technique is not suitable for all cases.
⇡ Removing a nasty squeak in Audacity
If you know the tools of sound editors well, you can eliminate interference using the sound editor tools in a comprehensive manner, and not just with one “noise reduction” filter. Suppose you are faced with another type of noise - a constant squeak at a certain frequency. This type of interference can occur, for example, due to strong interference from radio equipment. In this case, the principle of cutting out the frequency band that contains unwanted noise is used.
First of all, to do this you need to perform a frequency analysis of the track. To get such a graph in Audacity, run the command “Analysis → Build spectrum graph”.
Now carefully study the diagram - the frequency noise on the audio track will be visible as a protruding hump. Move your cursor over this area to find out the exact frequency of this peak. Remember this value. Now select the entire track and select “Effects → Nyquist Query” from the program menu.
In the window that appears, write something like this:
- (vector (notch2 (aref s 0) 6230 50) (notch2 (aref s 1) 6230 50)) for a stereo track,
- (notch2 s 6230 50) - if you have one audio track.
The number 6230 is the frequency at which interference is heard. The number 50 is a setting that you may need to experiment with to ensure that the cut frequency doesn't create too noticeable a dip in the frequency spectrum of the audio recording. You can verify that noise has been removed either by looking at the newly generated frequency spectrum or simply by ear.
⇡ Conclusion
Using high-quality equipment - expensive microphones, high-quality amplifiers, etc. - undoubtedly helps reduce noise. But in cases where it is impossible to influence the quality of the recording, it is worth remembering about digital audio processing. And even if it is not possible to completely get rid of third-party interference, in many cases the recording will become more pleasant to the ear. Therefore, if you set out to restore old recordings from a reel-to-reel or cassette recorder, be sure to try all possible options to eliminate noise, because, as you know, there is no limit to perfection.
Articles in the same category
For your attention, so that you don’t get bored, and even very, very useful, and completely free))) - How to remove background noise from a microphone? How to beautifully process your audio recordings? And make your voice like that of a central television announcer. And so, to your attention Adobe Audition.
How to remove background microphone noise
Articles on audio processing:
Sound, no one will argue with me, is the most important element for perception and assimilation of material. In our time, the Runet in this area practically does not think much about this topic, and the network often presents material that is sometimes simply terrible. Some people don’t think about this at all and their knowledge of sound consists only of buying a cheap microphone. They write videos with their comments, which sometimes, to put it mildly, you just don’t want to listen to. And to put it roughly, I want to spit.
So here's to those who are just starting out. And especially for those who have already entered the Internet with their products. Dedicated.
The simplest technique is noise reduction. Even if you don’t want to further improve the sound and make it beautiful and sonorous, this technique is enough.
At least it will suppress the recording noise of a low-quality microphone.
To do this, we will use the best, in my understanding, program, the audio processing program Adobe Audition CC. Decide for yourself how and where to get it. The most reliable way is to go to the official Adobe website using the link https://www.adobe.com/ru/creativecloud/catalog/desktop.html?promoid=KOVFF choose from the list and rent. Or another, if you consider it acceptable for yourself.
Noise suppression
First, let's listen to the original recording with natural noise that we have to remove:
Adobe Audition CC has a very good noise reduction program, which we will start working with.
The noise is gone! It's that simple!
Apply it. Write quality podcasts. Get rid of the noise. Make Runet cleaner!
Hi all! Today I want to show you how to very easily and simply remove noise from audio. Because any, even a very high-quality microphone, can record noise, if there is any.
This is in no way professional audio processing, it is just amateur noise removal that gives good results. I've been using this scheme for several years now and I'm happy with it.
But there is one more caveat, depending on what noise you are going to remove. If it is some kind of harsh hum, hissing, clattering, interference, and so on, then my method is unlikely to help you. Here it is better to take care of the room in which you record sound and possibly replace the microphone. I'll show you how to remove light noise.
I record sound in a program and remove noise in it. Therefore, I will show everything using the example of this program.
First of all, we record the sound. To do this, on the Recording tab, there is a Record button.
In principle, if you already have audio, you recorded it earlier in some other program, then you can easily open it in AVS Audio Editor. Simply drag and drop the audio file into this program.
Now we need to provide the program with a noise sample. To do this, press the button Get noise profile.
On our audio track we indicate to the program 2-3 seconds of noise. That is, we select an area where there is noise, but there is no conversation.
Select it and click OK.
Noise profile loaded. There are still some options to make here, but I usually leave everything at default. I said earlier that this is just amateur noise removal, so here we just click OK.
But you, in turn, keep in mind that there is Uninstall options. Experiment with these sliders as needed.
After a few seconds the noise will be removed.
It remains to check what happened.
If everything is ok, then right-click on the audio and select Save As.
We select the appropriate settings ( if the original audio was WAV, then select it) and save a copy of the already processed file.
How to successfully remove noise?
Next time you record audio, be sure to record the noise. When you start recording, be silent for a few seconds and let the microphone pick up any possible noise. Then, when removing noise, it will be easier. Because at the beginning of the audio we will already have silence, from which we will take a noise sample.
Audio check
And finally, I want to tell you to use headphones when removing noise. When you remove noise from the audio, test it with headphones before saving the track. Because it is with headphones that you will be convinced that there are no problems with sound.
It happened to me once that I chose a sample a little incorrectly and the audio came out with some artifacts. I checked without headphones and saved it as is. Then only during installation I discovered the jambs, but it was too late. I don't have any source codes left. I had to record the audio again! All you had to do was undo the changes using Ctr+Z and specify a different noise sample!
That's it. That's all, thanks for your attention. Bye everyone!
As you know, the sound recording quality of modern smartphones, tablets and laptops leaves much to be desired, to put it mildly. What to do if you need a high-quality audio recording, but all you have at your disposal is a track with buzzing, hissing and muttering? The answer is simple: install Audacity on Ubuntu and apply the wonderful Noise Removal filter, which will do all the work for you.
How to Install Audacity on Ubuntu
Apt-get install audacity
How to remove noise
Launch Audacity, open your audio file for cleaning. You will need to find a few seconds of silence in the recording. They are needed to create a noise profile of the recording. That is, silence means the absence of external sounds (for example, a pause between remarks). It is during these seconds that only the so-called white noise, created by the recording device and microphone itself, remains on the recording. It is this white noise that is needed so that the Noise Removal filter in Audacity can create a mathematical model and “subtract” extraneous sounds from the main stream.
Well, take a few seconds of silence and go to the Effect menu. There, find the Noise Removal filter. A window will appear asking you to create a noise profile. Feel free to click the Get Noise Profile button and follow the instructions. Then you can move on to the second cleaning step. You will be prompted to adjust several noise reduction settings. They are selected experimentally. Try different values, press the Preview button, listen and look for the optimal settings. When you are satisfied with the cleaning result, press the OK key. Ready.
Loading...
