I did a video interview for my job. I used a Bluetooth wireless microphone connected to my smartphone.
However, despite this system have worked super well on many occasions, I ended up with a big crackling.Unfortunately, I will no longer have the opportunity to do this interview again.
I have join a sample of the problem :
I tried the ânoise reductionâ function on Audacity but it seems that the crackle is not detected.
Do you have any solutions to solve this problem ? (if it can be solved âŚ)
Thanks for your help
PS : si jamais des français passent par ici, je suis preneuse dâune rĂŠponse dans ma langue !!
I suspect that crackling was data interference either from BT or USB.
You are not going to get all of those clicks out.
Have attached a cleaner version, any more processing and the voice is going to start sounding metallic.
Used two passes of gentle noise reduction, followed by some EQ (+3dB from 2 to 5KHz),
and finally, just a bit of compression (2:1 ratio with 50mS attack, 100mS release and about -10dB threshold).
I tried the ânoise reductionâ function on Audacity
I hope you have the interview as a WAV file and not MP3. Never do production in MP3. It creates distortion and it limits what you can do in processing.
That better?
First, I boosted the volume. Effect > Amplify > OK.
Then I applied the audiobook rumble filter. Effect > Filter Curve > Manage > Factory Presets > Low rolloff for speech > OK.
Noise Reduction works in two passes. Select the noise and then process it.
I drag-selected âcleanâ noise between 5 and 6 seconds.
Effect > Noise Reduction > Profile.
Then I applied Noise Reduction to the whole performance. Effect > Noise Reduction > 12, 6, 6 > OK.
Then I did the last two steps again two more times.
Profile > Reduction > Profile > Reduction.
I guessed at the first number, 12. You may get slightly higher voice quality with a lower correction such as 9, 6, 6.
That bubbly sound at the beginning is, I think, a problem with your MP3 file. You may have to clean those up manually.
I agree you can get that kind of interference from the phone being a phone. Use Airplane Mode to turn off the radio parts during recording. You can also get that kind of interference from the phone screen. The screen radiates trash and can get into sound production.
You can do experiments with all these to sort where your noise is coming from.
I have a stupid French joke. Everything sound better in French. âMy house is on fire and my car wonât start.â sounds wonderful in French.
9, 6, 6 seems to work OK. There is still a tiny bit of noise still there if you turn the volume up. So thatâs the limit.
I want to be clear that Iâm taking a new Noise Profile at each reduction pass to make sure Noise Reduction is processing the tiny damage left by the pass before.
finally, just a bit of compression (2:1 ratio with 50mS attack, 100mS release and about -10dB threshold).
Thatâs not part of damage control. Thatâs part of theatrical production mastering. That may not match your normal mastering and it locks out any other damage processes you may want to try. Thatâs why I picked Amplify at the beginning. Thatâs just turning the volume up. Thatâs reversible.
Thatâs not part of damage control. Thatâs part of theatrical production mastering. That may not match your normal mastering and it locks out any other damage processes you may want to try.
I beg to differ, itâs very much part of damage control.
If you get the correct compression ratio, it will actually help in making the dialog stand out more.
Same reason why I did EQ before the compression, which will help as well.
Multiband compression will also work, but doubt that the OP will have access to it as Audacity does not ship with one.
Of course if one over does it with compression, well, itâs a one way ticket, however, look at my final levels.
They are higher than is recommended for either broadcast or social media.
That was done deliberately so that some final compression can still be used on top of that.
A word on the Rumble Filter (Low Rolloff for Speech). Your crackling is not just high pitch ticking. The sound goes lower in pitch, too, and extends below 100Hz. Nobody is going to hear that, but that sound can affect the other tools.
That Audacity filter was designed to copy the filters that outside sound recording systems have to avoid wind noise and other low pitch distortions and rumble sounds.
The Zoom H1n portable recorder has three different ones depending on how bad things get.
Post back about what you decided to do and how it worked.
I think your first experience with Noise Reduction didnât have the Profile step included. Thatâs why it didnât do anything. Itâs pretty unusual for tools, filters, and effects to require two different steps, but Noise Reduction is one of those tools.
Paul2 didnât post which Noise Reduction values he usedâŚ
7, 6, 12 - but they are really only a guide.
The higher you go, especially on the first number, the more ânoiseâ will be reduced
however at a cost, the wanted audio starts sounding metallic.
That of course is subjective, the OP may accept more (or less) of the metallic sound than either of us.
Hence the term I used to quantify it, âgentleâ.
The OP may be very happy with perhaps something as high as 10,10,12.
On the other hand, he may opt for 3,6,12.
Furthermore, you seem to have misinterpreted my method of repair and categorically stated
that somehow yours is the only bona fide way.
There are often many paths to the same destination, just because the path you always take
works for you, does not mean that itâs either the best or only route.
The point is, experiment, as long as itâs on a copy.
This then is the only way to get to know the tools well and not just copy numbers and settings
without knowing exactly what they do and why.
You may even discover new, exciting things on your detour.
The default setting is setting is 3, settings lower than this tend to favor music and higher settings tend to favor spoken word.
andâŚ
âŚif artifacts remain in the noise-reduced audio the smoothing can make those artifacts sound more acceptable.
I find that Audacityâs NR is pretty dated and tends to be rather harsh, compared to other (newer) algorithms.
A setting of 10-12 for dialog helps for most applications.
Thanks a lot for all of your answers ! Iâll try the methods that you explain and choose the one that works the best for me. Hope it works as well as the sample you had send me
In fact, my sample was in .mp3 because the audio is an extract of a video⌠Wich last approximatively 7minutes so thereâs a lot of work if I must do this :
if I must
No, because we heard the crackling before the begining. So we make sure that everything was off⌠except my phone ! So Iâll try to use the airplane mode for the next time.
In fact I really did this 2 steps⌠but maybe not correctly !
