Is it possible to recover sound of the original voice after it has been changed by a voice changer?

What precautions can I take to prevent recovering the sound of original voice, do I have to build my own voice changer?

I'm considering this product: https://www.amazon.in/Yafa-Portable-Microphone-Broadcast-Disguiser/dp/B086K1XNTR/

I don't know if works by just shifting the pitch or uses something else. I want something which will make it difficult or impossible to recover original voice.

 

For what purpose? The ToS/UA prohibit discussions related to illegal activity, so on something like this you need to convince us that there is legitimate and legal intent.

 

For what purpose? The ToS/UA prohibit discussions related to illegal activity, so on something like this you need to convince us that there is legitimate and legal intent.

I want to talk about games on YouTube/Twitch but I don't my want my voice to be cloned and made to say things I might not say.

 

Definitely. Random noise is a better choice. Of course, you need an excellent algorithm.

Random Noise Software algorithm is the significant factor.

 

Speech to text —> text to speech.

 

@shell.albert

How will random noise generator help?

@ya'akov I've been trying to tag you but you don't seem to be available in the list of users which pops up after the "@" symbol. Why is that?

It will be too tedious to do that, I may have to sign up for some cloud service, I want to do this freely.

 

@shell.albert

How will random noise generator help?

@ya'akov I've been trying to tag you but you don't seem to be available in the list of users which pops up after the "@" symbol. Why is that?

It will be too tedious to do that, I may have to sign up for some cloud service, I want to do this freely.

Some small bug in the forum software, it should be sorted eventually.

By the way, although I know that you aren’t likely to want to hear it, I think your concern is out of scale with the reality of the situation. I really don’t believe anyone will be trying to fake your voice, there would be no profit in it. Not too be insulting, I have no intention of that—you aren’t important enough for someone to bother.

Additionally, when deep fakes become so common that even “unimportant” people are able to harass other “unimportant“ people with them, no one will trust Internet media (audio/video) as primary evidence anyway, so preventing it will not be so important. It will even become a defense for really gaffs and faux pas—“it’s fake (news)!”.

Once again, “unimportant” is not a value judgement, in this context it is intended to refer the socioeconomic resources (fame, money, clout, &c) controlled by a person. From a moral/ethical/humane perspective, you are no less important than someone who would be called “important” in the former context. I just needed some way to talk about it.

 

I think I should clarify: today, using computational voice cloning and faking is really simple. But you proposed to use a hardware voice changer. So, faking the voice changer would be simple but removing the effect of the changer is not simple right now, and the resources to do it are the trouble I was referring to.

One way to defeat it is to use a deep fake generator yourself, then your voice involved. But, to be clear, I wouldn’t trust it to hide your identity because speech patterns, vocabulary, and other semantic characteristics of your speech are enough to “fingerprint” you as yourself if there is any access to your real voice, etc.

 

These days hardware is becoming very powerful, soon you might have something, which may not even require training or is so well trained they may only need a sample of few seconds to clone. Like shown in this video:

One of my concern was my voice might cloned and made to say something which might be incriminating, or something which might be embarrassing.

I don't really know how the voice changer in the link works, does it change the pitch of the voice, if it does, it will be relatively easy to recover the original voice.

 

Thesimple scheme is to use a text to speech program like used for visually handicapped individuals. Since yur speech would never be part of the process there is no possible way to duplicate your speech. And that software is common and so even that would be hard to track back.

 

@shell.albert

How will random noise generator help?

No much. If you really need voice range signal protection you need to encrypt it to some extent or use Speech to text —> text to speech as posted by others.
We had a system called Parkhill as a voice scrambler.
https://wiki.radioreference.com/index.php/KY-65

 

Every day I get long distance phone calls from "little girls" in India selling "Duck Cleaning" and other things.
The callers have a very strong accent from India and hundreds of other people are heard talking in the call center.
They must be guys who are using the voice changer product linked here that is sold on Amazon in India.

Sure it simply changes the pitch. Then another one can change the pitch back to normal.

 

Thesimple scheme is to use a text to speech program like used for visually handicapped individuals. Since yur speech would never be part of the process there is no possible way to duplicate your speech. And that software is common and so even that would be hard to track back.

But as Ya’akov has said, audio is only one way of identifying you.

I wouldn’t trust it to hide your identity because speech patterns, vocabulary, and other semantic characteristics of your speech are enough to “fingerprint” you

 

selling "Duck Cleaning"

What did you expect with that avatar?

 

Some of the callers are selling medicare, others want to lower my credit card interest rate, Some want to sell me an extended auto warranty. And often I get hangup calls. What is needed is a way to prevent caller ID spoofing.

 

They say Duck because they can barely speak English and in their country have never seen a "duct".
"Hang up" calls are from immigrants who dial the wrong number and are surprised to hear me answer in English.
I do not know if a burglar phones me to hear if I am not home so he can break in and rob.

 

They say Duck because they can barely speak English and in their country have never seen a "duct".
"Hang up" calls are from immigrants who dial the wrong number and are surprised to hear me answer in English.
I do not know if a burglar phones me to hear if I am not home so he can break in and rob.

It is not reasonable that I would get several of that variety of calls every day. Nor would immigrants or wannabe burglars call that frequently.

 

@nsaspook If I use voice encryption, I doubt many people on YouTube & Twitch will understand what I'm saying. So there is no effective way to change voice which might prevent others from recovering the original voice.

Every day I get long distance phone calls from "little girls" in India selling "Duck Cleaning" and other things.
The callers have a very strong accent from India and hundreds of other people are heard talking in the call center.
They must be guys who are using the voice changer product linked here that is sold on Amazon in India.

Sure it simply changes the pitch. Then another one can change the pitch back to normal.

My main & only concern is my voice might be cloned and made to say embarrassing things. Nothing to do with masking accent, or anything. I'm an Indian and I think majority of people here who can speak in good English have neutral accent, which is neither American, British, Australian, etc. Neutral accent is probably the best way it should sound because it is lot more intelligible than other accents. In case you don't know the neutral accent I'm talking about, you can consider watching few Indian English news channels, like CNBC TV 18.

Here is a sample of what I mean by neutral accent:

This woman's accent is not entirely neutral, she has some idiosyncrasies of her own but it is very close to neutral and that is the only I could find without searching for hours.

 

You could always talk differently using a made up voice. It occurred to me long ago having an alternative voice and appearance could save my life one day in an emergency situation if my known identity was targeted. I got the idea from Rush Hour where the main character says "Sorry, I am from out of town".

I don't intend to defraud anyone but I have practiced several voices and tones mostly from a comical perspective. I could pass as a German or the like with very poor English. I have been defrauded myself in the past so I value being prepared towards those who see an easy target. I don't really see this as a problem though, more of a what-if consideration.

 

I would also like to add to not get caught up in what I like to call "internet wars". By this I mean people on the internet who only seek to argue or make a fool of you. Thier intentions are foul to begin with so as soon as you detect it, be done with them. If someone (a loser in my opinion) is going to record and change your voice to make fun of you, the best thing to do is rid yourself of that person. Let them make their recordings, why does it have to affect you if your online banking for example is secure?

I'm played many online video games in my day and sorry to say, most of the people in those chat rooms are fools with nothing better to do. Concern yourself with things of merit and don't worry about non-issues. They become issues if you let them, and a bully does not care who they harass. Don't let it be you.