ProVoice audio samples requested

[mitaux8030]

Hi all,
I'd like to request two fairly high quality audio samples of some ProVoice as heard on a communications receiver (taken from a discriminator output would be perfect)... but there is a catch.

I need one recording to be from group that is KNOWN to be unencrypted digital ProVoice
and
I need one recording from a group that is KNOWN to be encrypted digital ProVoice.

If you can help with one or the other - or both! - I'd be very appreciative.

What I'm trying to do is determine if it is possible to distinguish unencrypted ProVoice from encrypted ProVoice just by ear. FYI it's quite easy to distinguish MA-Com's earlier Aegis encrypted vs. unencrypted - the difference is subtle but easy to pick once you know what to listen for.

The ProVoice samples that live on the 'net already lack any information as to wether they're encrypted or not.

[Unitrunker]

It is VERY easy to tell if a ProVoice call is encrypted (but not by ear).

The header of every voice frame includes a 64 bit crypto IV (initialization vector). The "in-the-clear" crypto IV is the hexadecimal value 0xD2 repeated eight times.

I can send you samples of both. I can't tell them apart without looking at the actual bits.

[mitaux8030]

Did you get my PM Rick?
Ta - Jason

[Unitrunker]

Yes. The kids made me cook dinner. I'll dig up those samples.

[mitaux8030]

Still hanging out for those samples when you get the time... thanks!

[poltergeisty]

Have you seen this? http://wiki.radioreference.com/index...version_Method

[mitaux8030]

I have seen that, but not interested in that.
What I want, as per the first message, is two samples of provoice, one KNOWN to be encrypted, one KNOWN to be unencrypted. I want to see if I can pick the difference by ear alone. Just for my own interests sake. None of the samples on the web say wether they're encrypted or not, and they're not real great quality.

Anyone able to help? Anyone at all?

[car2back]

Post your request in the Oklahoma forum, someone in the Oklahoma City Area probably wouldn't mind providing some recordings of OCPD traffic for you!

[poltergeisty]

Originally Posted by mitaux8030

I have seen that, but not interested in that.
What I want, as per the first message, is two samples of provoice, one KNOWN to be encrypted, one KNOWN to be unencrypted. I want to see if I can pick the difference by ear alone. Just for my own interests sake. None of the samples on the web say wether they're encrypted or not, and they're not real great quality.

Anyone able to help? Anyone at all?

Could you hear the difference between an MP3 encoded in 128 VBR vs a MP3 encoded 128 Constant bit rate...?


VBR 128 Ssettings were, low 128/high 144

Constant bit rate 128



Well, I guess there is a difference.. I can hear it anyhow. The idea is behind the way the data is encode rather than how it is heard. In other words, can you hear the encrypted bits vs non encrypted bits? There would be no change in audio to the ear. Maybe to a dog. You need a certain hearing range...like a computer..

[js_scan888]

Originally Posted by poltergeisty

Could you hear the difference between an MP3 encoded in 128 VBR vs a MP3 encoded 128 Constant bit rate...?

can you hear the encrypted bits vs non encrypted bits? There would be no change in audio to the ear.

Think of it like this: A drummer/percussionist plays a solo from a sheet of music. To the trained ear you will hear the different beats, rhythm patterns, and knocking sounds being played. Now the musician scrambles/encrypts the notes on the music sheet and plays the solo again. You will notice that the patterns and beats you were hearing are now different. Listen to a data channel for several minutes and you will start hearing the various patterns. I would think that general voice data patterns would be different when encrypted but it also depends if the clear data is interleaved beforehand and what type of modulation is being used. The original poster is trying to find a reference point in order to do his own comparison.

[mitaux8030]

yes, I could actually (only just!) pick the difference audibly - there's less mp3 artifacts for the higher pitched components with the VBR recording.

Like I said, its quite easy to pick the difference with Aegis encrypted vs unencrypted, once you know what you're listening for. I'm wondering if provoice has that same trait?

Anyway I will pose the question to the Oklahoma section. That'll definitely be encrypted, yes?

[DCMon]

I know this is not what's being asked for but it lets us know what it sounds like anyway.
Google search turned up this site of audio samples and this one is reported to be the
OKC system: http://www.w2sjw.com/sounds/EDACS%20ProVoice.mp3

[Unitrunker]

Originally Posted by mitaux8030

Still hanging out for those samples when you get the time... thanks!

Hi Jason;

Check you PM box.

[mitaux8030]

Thanks UniTrunker, very much appreciated.

So there IS an audible difference. I can clearly pick it - its actually easier to pick than Aegis encrypted vs unencrypted. Curiosity sated at last!

(this comes from a guy who used to listen to C-64 'fast-load' tapes and try to pick the loading baud rate... yeah I know, wierd!)

[MattSR]

I have done the same with clear and encrypted P25 too.. there are subtle differences that a trained (crazy?) person can distinguish the two

[Unitrunker]

Originally Posted by mitaux8030

So there IS an audible difference. I can clearly pick it - its actually easier to pick than Aegis encrypted vs unencrypted. Curiosity sated at last!

The "difference" you hear may be due to the content (eg. what was actually spoken) rather than the use of a different key.

For a real test - you would need the same audio - encrypted three different ways (including one clear sample).

[mitaux8030]

Stictly speaking that would be correct. But once you consider that encryption is effectively going to remove any trace of 'regularity' to the resultant bit stream, you can then say with reasonably high confidence that neither content nor key will (or should, if the designer is worth his salt!) be a factor in this.

I believe what I'm hearing in the unencrypted sample is regular vocoder framing patterns - and these are largely absent in the encrypted sample. Any trace of 'regularity' in the encrypted sample would be Layer 1 like Um framing. I've observed the very same effect with Aegis too.

Anyway, I'm not going to get too serious or categoric in my findings, just a 'this is the most likely situation' is good enough to satisfy my curiosity. If I were seriously trying to reverse engineer the samples then I'd have to go to that level of course - but that's not what I'm doing. Far from it, I have only very minimal training in cryptology, I definitely wouldn't have the skills to do anything like this.

[RiceCake]

The biggest goal in cryptography is to remove any "noticeable" marks in the stream in relation to the data. Any algorithm that can be predicted can be broken without knowing the key. The harder the key and the harder the algorithm, the more scrambling, or "rounds", the data goes through...

In other words, lets take the phrase...

The quick brown fox.

In unencrypted form over a radio system, we can (lazily) use numbers to denote what this should sound like when spoken by using higher numbers for higher pitches...

00012300030021002303031110001233322

Bear with me...the zero's are when nothing is spoken, 1 is a low tone, 3 is high, etc... In actual digital communications, these bits fly by at much faster (9600 bps on some) speeds to allow way higher fidelity, so it might be hard, by ear, to even remotely pick up these changes.

But, here comes cryptography. Fun!

Lets say the number 1 is our encryption key. Now, adding 1 to every number is kinda pointless. In our super-primitive digital system, it would only make the pitch higher. Lets have some more fun, we'll add 1 to every second number, and every other number we'll add the number to its left to it.

11226411441142116444443221114374632

Very rudementary? Yes. But in terms of what digital sounds like?

It should sound exactly like pure static, regardless of what people say on it.

[greeneyes965]

Why do the digital radios have too be encrypted

[RiceCake]

The same reason why I wouldn't post my credit card number on this forum.

You have no idea who can be listening.

Encryption fixes that.