I'm thinking about writing a fairly simple voice encryption system. I plan to digitize the voice signal with an ADC, do some simple DSP, then pump the result through a DAC and then transmit.
Does anyone have any experience with this? I have an idea of how I might go about it, but I wanted some input to see if I'm on the right track. I don't want to post my idea just yet, because I don't want to contaminate your minds with my idea.
Any thoughts are appreciated.
Voice Encryption
-
HighFrequency
- Posts: 122
- Joined: Sun Apr 17, 2005 1:01 am
- Location: Victoria BC
- Contact:
Voice Encryption
There is only one correct answer, mine.
well, you didnt tell to much what you want to do with the DSP...
Why do you need a DSP?
Here is a sample project from TI.. i think everyone will soon call me Mr. TI
http://focus.ti.com/lit/an/slaa123/slaa123.pdf
I made this project a while ago..its a Solid State Voice Recorder Using Flash MSP430. If this project interests you here is where you can find the code for the micro..
http://focus.ti.com/docs/apps/catalog/r ... me=slaa123
If you plan on doing some advanced filtering.. maybe start out with a dsp.
Why do you need a DSP?
Here is a sample project from TI.. i think everyone will soon call me Mr. TI
http://focus.ti.com/lit/an/slaa123/slaa123.pdf
I made this project a while ago..its a Solid State Voice Recorder Using Flash MSP430. If this project interests you here is where you can find the code for the micro..
http://focus.ti.com/docs/apps/catalog/r ... me=slaa123
If you plan on doing some advanced filtering.. maybe start out with a dsp.
-
dyarker
- Posts: 1920
- Joined: Fri Aug 22, 2003 1:01 am
- Location: Izmir, Turkiye; from Rochester, NY
- Contact:
Real time (like telephone call), or recording and store?
"... simple DSP, ..." sounds more like scrambling than encryption. Voice encryption would be analog to digital then encrypt with DES, AES, Arc Four, etc).
Which CODEC (ADC/DAC format) you planning on? (uLaw, CVSD, LPC, etc) If real time, the amount of bandwidth available affects choise of CODEC. uLaw needs 64Kbps. CVSD has good voice quality at 16Kbps, usable at 8Kbps. Some CODECs have usable quality as low as 2400bps, but the ADC takes more processing.
Not contaminating the idea well is a good idea, but we need a little more info.
Cheers,
"... simple DSP, ..." sounds more like scrambling than encryption. Voice encryption would be analog to digital then encrypt with DES, AES, Arc Four, etc).
Which CODEC (ADC/DAC format) you planning on? (uLaw, CVSD, LPC, etc) If real time, the amount of bandwidth available affects choise of CODEC. uLaw needs 64Kbps. CVSD has good voice quality at 16Kbps, usable at 8Kbps. Some CODECs have usable quality as low as 2400bps, but the ADC takes more processing.
Not contaminating the idea well is a good idea, but we need a little more info.
Cheers,
Dale Y
-
HighFrequency
- Posts: 122
- Joined: Sun Apr 17, 2005 1:01 am
- Location: Victoria BC
- Contact:
The idea is a crypto device for a 2-way VHF radio. The reason I want the transmission in analog is for bandwidth, and I'm pretty sure you can't send real-time voice digitally.
I want the DSP to do the encrypting. I've seen some analog scramblers, but that's not really what I'm looking for.
I want the DSP to do the encrypting. I've seen some analog scramblers, but that's not really what I'm looking for.
There is only one correct answer, mine.
It's not immediately apparent to me how the receiver could decrypt an analog signal produced with a digital transform at the sender. Maybe.
A simple analog style encryption is to invert the audio basedband at the transmitter using all analog techniques and a mixer producing sum and difference products as all mixers do. These two products are inverted from one another. The inverted one is unintelligible to a passive listener.
A simple inverse of this on the receiving end "decrypts".
You could do this with a DSP that's fast enough to sample audio at the Nyquist rate or more, then do this inversion, and vice versa on the other end. A lot harder than an all digital system.
A simple analog style encryption is to invert the audio basedband at the transmitter using all analog techniques and a mixer producing sum and difference products as all mixers do. These two products are inverted from one another. The inverted one is unintelligible to a passive listener.
A simple inverse of this on the receiving end "decrypts".
You could do this with a DSP that's fast enough to sample audio at the Nyquist rate or more, then do this inversion, and vice versa on the other end. A lot harder than an all digital system.
-
HighFrequency
- Posts: 122
- Joined: Sun Apr 17, 2005 1:01 am
- Location: Victoria BC
- Contact:
dyarker... I'm pretty sure the military does it.
stevech... I agree it won't be easy. That's what makes the project difficult (and thus interesting). At the receiver I would simply amplify the signal to the correct level, sample it at the same rate as during transmission, pass it through a DSP and then through a DAC and out to the speaker or whatever output device I decide to use.
stevech... I agree it won't be easy. That's what makes the project difficult (and thus interesting). At the receiver I would simply amplify the signal to the correct level, sample it at the same rate as during transmission, pass it through a DSP and then through a DAC and out to the speaker or whatever output device I decide to use.
There is only one correct answer, mine.
When you're talking about encryption, you have to first access what the immediate and long-term value of the payload is.
There isn't an encryption that will never be broken... therefore you have to decide,
Do you simply want to make it difficult for the average person to hear,
Do you want it impossible to decrypt in real time but ok if it takes 2 hours to decrypt? If not 2 hours, what is acceptable, a week, a year, 5 years?
For example, I have friends working as missionaries in locations that are very hostile to Christian missionaries. Communication is done encrypted using as strong of an encryption capability as possible; something that would take 5 years or more to decrypt. People can die if they're found out.
How much do you need?
Also, do you need public key encryption, and if so, how do you plan to exhcnage keys?
If you simply need to make it difficult to hear, use something simple like the common xor technique. Just about any processor would be able to encode and decode that quick enough to be convienent. It's not technically encryption, but you can get plenty of details by searching for xor encryption.
You could also combine something simple like xor with steganography to make it less trivial to decode.
If you need something really strong like aes, you may not be able to do it without special controllers. I did a quick search and there are special DSPs with aes cores for this type of thing.
There isn't an encryption that will never be broken... therefore you have to decide,
Do you simply want to make it difficult for the average person to hear,
Do you want it impossible to decrypt in real time but ok if it takes 2 hours to decrypt? If not 2 hours, what is acceptable, a week, a year, 5 years?
For example, I have friends working as missionaries in locations that are very hostile to Christian missionaries. Communication is done encrypted using as strong of an encryption capability as possible; something that would take 5 years or more to decrypt. People can die if they're found out.
How much do you need?
Also, do you need public key encryption, and if so, how do you plan to exhcnage keys?
If you simply need to make it difficult to hear, use something simple like the common xor technique. Just about any processor would be able to encode and decode that quick enough to be convienent. It's not technically encryption, but you can get plenty of details by searching for xor encryption.
You could also combine something simple like xor with steganography to make it less trivial to decode.
If you need something really strong like aes, you may not be able to do it without special controllers. I did a quick search and there are special DSPs with aes cores for this type of thing.
-
HighFrequency
- Posts: 122
- Joined: Sun Apr 17, 2005 1:01 am
- Location: Victoria BC
- Contact:
Well for now, it's for a school project. At this point, I'm not too concerned with the strength of the encryption. I'm really just looking for something that's a little stronger than just inverting the modulating signal, but that is relatively easy to decrypt at the receiver.
My plan is the XOR method. I just wanted to see if anyone else out there had done this sort of thing, and if so, what method they used.
My plan is the XOR method. I just wanted to see if anyone else out there had done this sort of thing, and if so, what method they used.
There is only one correct answer, mine.
-
dyarker
- Posts: 1920
- Joined: Fri Aug 22, 2003 1:01 am
- Location: Izmir, Turkiye; from Rochester, NY
- Contact:
XOR of voice bit stream with pseudoramdom key stream is encryption. When you said in the DSP, that led me the wrong direction. Just messing up the signal, analog or digital, would be scrambling. Pardon.
When you DAC the encypted bytes to send to the transmiter as analog, the bandwidth will be much wider, with wild shifts in instantaeous frequency. Any outside the RF bandwidth of the VHF channel will be lost. So at the receiver, the result of incoming cypher stream XOR with key not stream will be complete rubbish. The speaker output will be like white noise modulated by skweaks(sp?) and squawks. That's why stevech said "... and they always use a digital data stream ...".
If at the transmitter you use steep low pass filter at 2.9kHz on the analog (in DSP), ADC it using CVSD at 6KHz, then reduce the amplitude of the digital signal and feed that to the transmit FM modulator; you might have an FSK VHF signal that will pass. At receiver, amplify, schmitt trigger the "audio", decrypt, and CVSD to analog.
------------------------------------------------------------
Yes, the military (and police) have encrypted VHF radios. What they can do on their assigned channels has nothing to do with what's legal on ham and industrial channels. They use digital radios, the RF bandwith is more twice that of a analog voice radio. The FCC considers that when assigning frequencies. If you transmit a wider than approved signal, you will interfere with next higher, and lower, channels. Someone will complain and the FCC will come and take you away ha ha.
Cheers and good luck with the project,
When you DAC the encypted bytes to send to the transmiter as analog, the bandwidth will be much wider, with wild shifts in instantaeous frequency. Any outside the RF bandwidth of the VHF channel will be lost. So at the receiver, the result of incoming cypher stream XOR with key not stream will be complete rubbish. The speaker output will be like white noise modulated by skweaks(sp?) and squawks. That's why stevech said "... and they always use a digital data stream ...".
If at the transmitter you use steep low pass filter at 2.9kHz on the analog (in DSP), ADC it using CVSD at 6KHz, then reduce the amplitude of the digital signal and feed that to the transmit FM modulator; you might have an FSK VHF signal that will pass. At receiver, amplify, schmitt trigger the "audio", decrypt, and CVSD to analog.
------------------------------------------------------------
Yes, the military (and police) have encrypted VHF radios. What they can do on their assigned channels has nothing to do with what's legal on ham and industrial channels. They use digital radios, the RF bandwith is more twice that of a analog voice radio. The FCC considers that when assigning frequencies. If you transmit a wider than approved signal, you will interfere with next higher, and lower, channels. Someone will complain and the FCC will come and take you away ha ha.
Cheers and good luck with the project,
Dale Y
-
HighFrequency
- Posts: 122
- Joined: Sun Apr 17, 2005 1:01 am
- Location: Victoria BC
- Contact:
No, not really.
DSP is digital signal processing, usually with special fast hard math chips, not just sampling.
FFT is Fast Fourier Transform. By magic and advanced math it just works.
FFT is used very often in DSP. It is hard to be doing one without even knowing the other.
What it does is take a wave and separate it into all the frequencies it contains.
Before, the bins contain the volume of the wave in time order.
After, the bins contain the volume of all the frequencies in frequency order.
A graphic equalizer display on a sound system is a good example of what an FFT does.
DSP is digital signal processing, usually with special fast hard math chips, not just sampling.
FFT is Fast Fourier Transform. By magic and advanced math it just works.
FFT is used very often in DSP. It is hard to be doing one without even knowing the other.
What it does is take a wave and separate it into all the frequencies it contains.
Before, the bins contain the volume of the wave in time order.
After, the bins contain the volume of all the frequencies in frequency order.
A graphic equalizer display on a sound system is a good example of what an FFT does.
Who is online
Users browsing this forum: No registered users and 0 guests