How to blur sound?

If I have a great photo, but there is something in it I don't want to show, I can often blur it out and it's great.  A face, a licence plate, whatever.  

This week I recorded some sounds from a place.  There are great sounds, but sometimes the sounds are quieter and you can hear music.  

Music that I will get in trouble if I post on youtube.

I want to make the sound of that music blurry - but for hearing?
...but let the main part of the sound come through.  The main part are louder sounds, so I just need to 'blur' the quieter moments.

What do I need to know to be able to do this?  (on a really crappy computer that can't even handle audacity)?  

Is this possible?  

The answer depends on many factors. If it is a stereo recording, you can sometimes play with the phase to null or obscure certain elements. You can sometimes notch out parts of the eq spectrum that prominently feature the unwanted sound. Many times, the best way is to choose sections of the recording where the unwanted element is minimized/not present, and possibly loop it during other parts.  Then there are proprietary noise reduction plugins, like those from Wavelab, which can be effective, but are expensive and not user friendly to the novice. If you can describe the source file and the desired outcome, or post a sample, I could possibly comment further.

You may also find a program such as RX-9 by Izotope to do what you want.  They have a free 30 day trial , so the price can't be beat.  If memory serves, it has some presets to automatically filter out certain background noises.  Good luck.

https://www.izotope.com/en/products/rx.html

I don't know anything. I have a tin ear so I can enjoy more music than someone with perfect pitch. But I have an inquiring mind that recognizes no barriers and as an example will contemplate the Multiplication Of Loaves crossed with The Banach Tarski Paradox. I can also explain the Trinity.

So here is just my two cents: if stereo recording, play with two speakers and re-record. Play around with the balance. Try again with the wires to one speaker connected in reverse{phase cancellation}. Does that help? If not we enter mono territory - - -  how about playing another piece of music and rerecording the lot. If the music is mangled, will they pursue copyright claims? I have a tin ear so I can't help. But I have played around with SOX so I have some locus standi here. If the music bits are interrupted by SOX silence bits, will copyright claims hold water if you show the judge that you have tried? SOX can do low-pass, high-pass, band-pass filtering amongst other sound effects like reverberation and echo.  I have a few SOX settings I use like:

where to get SOX           {there is a PDF user guide}

sox z2.wav z4.wav norm -i contrast loudness {to enhance speech}
sox allah.wav al2.wav gain -6 treble +10 1000 bandpass 2000 3000 trim 00:10:20 1:11:25 {start duration}

sox                     2.wav   -s     22.wav

? sox -t alsa default ./recording.flac silence 1 0.1 5% 1 1.0 5%

SoX recording audio and trimming silence

sox in.wav out6.wav silence -l 1 0.1 1% -1 2.0 1%

useful SOX documentation

sox -n -r 44100 -c 2 silence.wav trim 0.0 3.0  {Generate 3 seconds of (stereo) silence}

sox                     2.wav   -s     22.wav
14:13 c:\addons\sox1420> sox Pater_Noster.wav  null.wav  stat -v
14:13 c:\addons\sox1420> sox GLoria_Patri.wav  null.wav  stat -v

14:24 c:\addons\sox1420> sox GLoria_Patri.wav GLoria_Patri_norm.wav norm

14:25 c:\addons\sox1420> sox GLoria_Patri_norm.wav  null.wav  stat -v
16:28 c:\addons\sox1420> sox GLoria_Patri_norm.wav  gpxx.wav  vol 2.3db
sox effects: vol clipped 6 samples; decrease volume?
sox sox: gpxx.wav: output clipped 6 samples; decrease volume?

You can run through the effects individually or mix the whole lot together.

I use QMC {quick media converter } to extract/convert audio to WAV, process with SOX
and use LAME to convert WAV to MP3.

Sound engineering is tough and tedious. Hats off to them. They deserve their wages.

BTW, this works on a 32-bit Windows 7 and also worked on 32-bit XP.

Audacity is an open source free download with numerous features which may help.

There has been recent progress in this area and a few public demos are available:

https://songdonkey.ai/ "Extract vocals and instruments with artificial intelligence"
https://splitter.ai/

I think I've asked the question wrong or I'm reading wrong.  Audio editing isn't something I understand well, so I'm guessing it's the latter.

I don't need to remove the song from the audio.  I like that it's a big part of the noise.  What I need to do is to make it less recognizable as a specific song - to avoid violating the copyright of the makers of that song.  

There is about 3 seconds here and there, where the big noises fade enough for the radio to come through with a few recognizable bits.  

If this was an image, I would take a soft brush and smudge out the colours so they blend together just enough to still be there, but not recognizable.  But I don't have the words for audio.  

Can you try layering more audio over the parts to blur? The human ear can't generally differentiate between 3 simultaneous sounds, so perhaps stack the song bit over itself, with a split second delay, a few times - or layer the various instances over each other. Will probably require fiddling to get it to sound good.

r ranson wrote:I think I've asked the question wrong or I'm reading wrong.  Audio editing isn't something I understand well, so I'm guessing it's the latter.

I don't need to remove the song from the audio.  I like that it's a big part of the noise.  What I need to do is to make it less recognizable as a specific song - to avoid violating the copyright of the makers of that song.  

There is about 3 seconds here and there, where the big noises fade enough for the radio to come through with a few recognizable bits.  

If this was an image, I would take a soft brush and smudge out the colours so they blend together just enough to still be there, but not recognizable.  But I don't have the words for audio.  

The replies are telling you how to do just that. The other way I can think of is you either fade out those bits where you can hear it, or add sound over the top of it. Because the sound you have recorded is a single file (only one microphone) you cannot easily separate the music from everything else.
The other way to look at what you want is how would it sound? Have you ever heard music where you cannot make out the song? have you ever come across a video where the music is anything other than faded out or covered up? Perhaps if you can find an example someone can say how it was done?

I think adding the song back over the top or trying to "blurr" it would make a horrible discordant cacophony.

We can also approach this from a copyright law/intellectual propertt perspective. There is such thing as "fair use," usually characterized by short lengths, and not profitting from the property of others. Again, "it depends" on many factors, and I'd have to hear it and know the context of the final form to advise.

i watch a lot of videos where people are walking in cities and pass buskers. They cut the sound when they pass the buskers, like literally dead air til they pass, saying that youtube has gone after them to pay rights even when someone else is playing music on the street.
In one case the person is quite tech savvy, and edits their videos pretty well, and either they're trying to make a point or, i think it's more likely it's just too much of a PITA to do anything more involved than to simply mute for a few seconds.

I uploaded it to youtube as unlisted and it looks like their automatic check didn't find any copyright violations - AT THIS TIME.  It doesn't mean it won't find more later.

stack the song bit over itself, with a split second delay,

That sounds like a good idea.   I think it would get the 'blur' sound I'm going for and I can see how to do that with my current software.  

If the sound is isolated, you can indeed just mute it..

I was assuming there are other parts of the sound you want to keep. Then the 5 channels of splitter.ai should give you a few options to remove the music without completely muting everything.

I'm not a fan of relying on Fair Use if I don't have to.

Mostly because I don't live in a country with Fair Use.  We have Fair Dealings which is far more limiting.  

And a massive list of reasons I want to stay as far away from Fair Use unless it's vital to the story of the video.

Tereza Okava wrote:i watch a lot of videos where people are walking in cities and pass buskers. They cut the sound when they pass the buskers, like literally dead air til they pass, saying that youtube has gone after them to pay rights even when someone else is playing music on the street.

The idea for that section is to show how noisy the sound is at that place.  I didn't realize they had music playing until I got home and reviewed the footage.  

The cutting of the sound is usually caused by someone making a claim or strike on the channel for using the music without a licence.  It's mostly done after uploading - sometimes by youtube automatically.

r ranson wrote:    
There is about 3 seconds here and there, where the big noises fade enough for the radio to come through with a few recognizable bits.  
 

If it is just a few bits, why not rerecord the audio and mute or turn down the volume during those bits? You mention Youtube so I guess there is video that goes along with it.

Try asking one of the boots to give this a go.

I don't want to reduce the volume.   The point is to show how noisy it is.   Muting or lowering would have the opposite effect.

Think like a picture.   If we blur a face we don't remove colour, we keep the colours the same intensity and value, but smudge them together a bit so the shapes are indistinct.

I have heard this done in professional production, where they make a song sound blury, but not youtube.

Ill try doing the out of time layering thing tomorrow and hear if that can do it.

One challenge here is to try to audiolize "blur". We can do this as an abstract thought process, but a visual blur is not something that is inherently possible in audio. It's essentially adding "noise". Noise will sound like static of some type. Alternatively ducking or muting volume is an option as mentioned, but you say it's not satisfactory. Stacking the audio won't trick AI trackers. I actually doubt adding noise will either. Those machine learning algorithms devised by IP holders are impressive and forever improving as long as it is within their best interest to improve them.

There are a lot of cloud software suites these days, I was impressed with the plugins offered on google drive for audio recording work.

Still. Audio editing is not simple.

I speak from experience doing phonetic audio analysis for linguistic studies.

Honestly I would recommend re-recording and hunting for IP free material in the recording.

r ranson wrote:
I have heard this done in professional production, where they make a song sound blury, but not youtube.

I'm curious as to the particular effect. Did it sound like it was underwater? That's probably a very particular effect or combination in some audio editing suite. I also doubt it will trick a modern AI.

Again, lack of language on my part.

Sometimes they use the underwater effect, but you can usually make out the song.

What I'm thinking it sounds like is somehow they used interference to knock off the big peaks and troughs to reduce the baseline and bring up the ambient noise enough to make up for the volume lost.  Sort of like how two sine waves can cancel each other out or how tinnitus machines reduce the scream in the ear by matching the frequency.  

But I suspect this takes massive skills and equipment.  

I sampled:
Copying and layering the sound 5 frames off did what I needed for this situation.  Good enough for the storyline.  

Thanks all.

My 2 cents...
You could "re-record" said video with another device.
Meaning, play what you want to show on screen (computer, laptop, etc.) while holding  another recording device (video camera etc.). That way you can adjust the volume on the original as you go, while re-recording on the other device.
I know, it doesn't sound hi tech but to me, it would be the easiest thing to do