Music as crypto medium

[Mok-Kong Shen]

Hi,

Music as a crypto medium goes back presumably almost to the
very beginning of the history of crypto. Nevertheless I found
it interesting to read a recent article written by a school
girl (C. Licht, in Junge Wissenschaften) on her own experiment
to use music to transmit secret informations. As far as I
(an ignorant in music) could summerize, she determined from
a number of works of Bach, Vivaldi and Mozart a frequency
distribution of the musical notes therein and used that to fit
in a rough manner the frequency distribution of letters in
a natural language, so as to obtain a mapping between the
two. Evidently such transcriptions couldn't be expected to
result in excellent pieces of music. However I suppose one
should perhaps even in the modern time not entirely neglect
the potential utility of music for crypto (including stego).
I should appreciate to know some more modern literature
references to papers relating music and crypto.

Thanks,

M. K. Shen
---------------------------------------------------------------

Was sich ueberhaupt sagen laesst, laesst sich klar sagen;
und wovon man nicht sprechen kann, darueber muss man schweigen.

L. Wittgenstein

[Globemaker]

On Sep 28, 7:14 am, Mok-Kong Shen wrote:
> ...transcriptions couldn't be expected to

> result in excellent pieces of music. However I suppose one
> should perhaps even in the modern time not entirely neglect
> the potential utility of music for crypto (including stego).
> I should appreciate to know some more modern literature
> references to papers relating music and crypto.

The Musical Instrument Digital Interface MIDI is useful for stego:
http://www.midi.org/
Not only notes are sent, side-channel messages are sent on midi serial
streams.
I do not have a crypto-midi reference for Mok-Kong Shen, but look at
the gamma ray use:
http://www.midi.org/aboutmidi/othermidi.php
One plausible ruse would be to profess to be an Arnold Schoenberg
afficionado and produce atonal sequences as music and as crypto.
Arnold "couldn't be expected to result in excellent pieces of music"
either, but it is complicated.

Alan Folmsbee, Kona, Hawaii

[mosherubin]

Cryptologia published a landmark article by Eric Sams in its October
1979 issue:

MUSICAL CRYPTOGRAPHY
Author: Eric Sams
DOI: 10.1080/0161-117991854052
Publication Frequency: 4 issues per year
Published in: journal Cryptologia, Volume 3, Issue 4 October 1979 ,
pages 193 - 201

You can order a reprint from Taylor & Francis, the current publishers
of Cryptologia:

http://www.informaworld.com/smpp/content~db=all~content=a741902805?words=eric|sams&hash=2154869284

Hope this helps,

Moshe

[Mok-Kong Shen]

I like to thank Globemaker and mosherubin for informations and
comments.

I read the paper by Sams. It is explained there how some well-known
composers enciphered names into their work. In general, no consideration
seemed to be given to the frequency distribution of letters. A, B, C,
D, for example, were paired with increasingly higher notes. There
was no necessity of doing otherweise, of course, since only a few
selected letters of the alphabet were used and only a few names
were enciphered. For transforming texts in general, however, the
technique employed by the school girl I cited could apparently be
an improvement, since there is a somewhat better fit to the distribution
of notes that naturally occur in musical works. (Mapping some letters
to more than one notes in the way of homophone cipher might improve
the matter still a little bit, I guess.)

Below are some other tiny ignorant's opinions of mine that I am
taking the liberty to express on this occasion.

In the nineteen sixties there appeared already the first machine
composed piece of music, which was premiered during an IFIP congress,
if my memory was right. Quite recently I saw papers dealing with
such topics as employing machine to study the styles of composers.
On the other hand, the technical advances in automatic generating
and recording/analysing music seem to be far reaching. This mean
that it may not too unrealistic today to expect that messages
could be encoded, either by software alone or by software with
human interactions, into music in such a way that it is hard to be
discerned by the uninitiated.

I am told that not all professional musicians play "exactly" the
same on one and the same work. Further, there are good and bad
players, the beginners and the experienced. The difference of
quality of performance is sometimes easily detectable even by
the laymen. This means that it would be in principle possible to
deliberately introduce some "noise" (the content of an encrypted
message) into an otherwise nice professional performance, the
result being then simply equivalent to a performance from a poorer
human player. Of course, how the receiver is to extract the
noise might not be a technically trivial matter.

Thanks,

M. K. Shen

[rossum]

On Mon, 28 Sep 2009 19:14:02 +0200, Mok-Kong Shen
<mok-ko...@t-online.de> wrote:

>Hi,
>
>Music as a crypto medium goes back presumably almost to the
>very beginning of the history of crypto. Nevertheless I found
>it interesting to read a recent article written by a school
>girl (C. Licht, in Junge Wissenschaften) on her own experiment
>to use music to transmit secret informations. As far as I
>(an ignorant in music) could summerize, she determined from
>a number of works of Bach, Vivaldi and Mozart a frequency
>distribution of the musical notes therein and used that to fit
>in a rough manner the frequency distribution of letters in
>a natural language, so as to obtain a mapping between the
>two. Evidently such transcriptions couldn't be expected to
>result in excellent pieces of music. However I suppose one
>should perhaps even in the modern time not entirely neglect
>the potential utility of music for crypto (including stego).
>I should appreciate to know some more modern literature
>references to papers relating music and crypto.
>
>Thanks,
>
>M. K. Shen

Hardly modern, but you might want to have a look at Mozart's
Musikalisches Würfelspiel:

http://sunsite.univie.ac.at/Mozart/dice/

Which has obvious possibilities for incorporating hidden information
into a piece of music.

rossum

[Milen Rangelov]

I remember trying to use a list of mp3 files as one-time pad. Turned
out to be a very bad idea - mp3 files contained long sequences of null
bytes. XOR-ing them with the plaintext generated large parts of
ciphertext equal to the plaintext. I guess mp3-compressed music is
actually a bad random source :)

Besides, it is not exactly "one-time" unless there is some way to
safely destroy the device (ipod) containing the mp3 playlist. Videos
like that one gave me some insight though :)
http://www.youtube.com/watch?v=jzFf3DkFcWs

I don't think anyone can derive even a single bit out of a blended
ipod. I thought that secretly exchanging ipods was kinda conspirative,
do not raise much of suspicion, etc

What a bad idea :)

MP3-based steganography must be a better idea :)

[Mok-Kong Shen]

Milen Rangelov wrote:

[snip]

> MP3-based steganography must be a better idea :)

I have no knowledge of the organization/Format of MP3 files at
all. If it were possible to insert into a MP3 file bit sequences
that have or almost have no practical consequences (much like
introducing dummy sequences in an exe file), the stego could
certainly be effectively done, I would think.

Thanks,

M. K. Shen

[Milen Rangelov]

In fact, I've done some experiments some time ago. Basically that's
possible and not so hard. You've got to avoid overwriting the ID3
headers (generally, avoid overwriting the first kilobytes of the file
is enough). Then the file consists of sequences of data headers
followed by variable-length data frames. Data headers are easily
recognized because they start with unique sync word and have a fixed
size.

In effect, if you avoid overwriting the headers and id3 frame, the mp3
stream tolerates lots of *modifications* without noticeable quality
degradation. Of course, the quality will start to degrade noticeably
once you modify too much data.

If you happen to overwrite a header then there is almost always a
noticeable degradation in quality (you hear unpleasant sharp and loud
noise or some short period of no sound at all at some position of the
output stream).

Generally, it is easy to conceal data in a mp3 file, because there is
no CRC mechanism to validate if the mp3 data frames were *damaged*.
But even if it had one, it wouldn't be so hard to calculate the
checksum of the modified data frame.

[Mok-Kong Shen]

I like to thank rossum and Rangelov for providing interesting and
valuable informations.

It seems to me that stego with sound files may have advantages over
stego with image files. For with the modern camera techniques,
pictures have barely random defects and so modifications would be
subject to easier detection, while sound recordings (excepting ones
that are done in special studios for the production of commercial CDs)
contain in general always some ambient noise, which may render the
decision between presence of stego or non-presence rather difficult.
Further, a picture can be conveniently repeatedly scrutinized on
a display, while doing the analogue for a sound record appears to
be less comfortable.

A bit OT: I presume that Pdf, ps, Latex etc. files could also be
modified to contain stego without too much difficulty. Exchange of
all the above mentioned file types should as a rule cause less
attention from the opponent then exe files, for they are normally
frequently exchanged by common people and form thus nice busy
channels [1].

I would be fine, if sometime there would be good freeware or shareware
available that automatically perform such stego with no or little user
interactions.

Thanks,

M. K. Shen
---------------------------------------------

[1] Ones that keep the opponent constantly "busy"! I am taking the
liberty here also to mention once again a humble (unfortunately)
very low bit rate busy channel that I suggested previously: Adapt
the word counts of lines of e-mails (with the exception of the last
line of any paragraph) modulo 2 to the information bits to be
transmitted.

[Mok-Kong Shen]

I got elsewhere a link to a page of a little bit related interest
entitled 'Music created by learning computer getting better':

http://singularityhub.com/2009/10/09/music-created-by-learning-computer-getting-better/

M. K. Shen

[Mok-Kong Shen]

Mok-Kong Shen wrote:
>
[snip]

> A bit OT: I presume that Pdf, ps, Latex etc. files could also be

> modified to contain stego without too much difficulty. .......

There is a recent paper concerning stego with PDF files:

I.-S. Lee, W.-H. Tsai, A new approach to covert communication via
PDF files. Signal Processing, vol.90, Feb. 2010.

(There is no typo above with '2010'. I had the hard copies of the
Jan. and the Feb. 2010 issue of the journal in my hand.)

M. K. Shen

[Mok-Kong Shen]

Milen Rangelov wrote:

> MP3-based steganography must be a better idea :)

For those interested in MP3: There is now a product MXP4, which seems
to be technically an advancement over MP3. See http://mxp4.com/

M. K. Shen

[Mok-Kong Shen]

Mok-Kong Shen wrote:
[snip]

> It seems to me that stego with sound files may have advantages over
> stego with image files. For with the modern camera techniques,
> pictures have barely random defects and so modifications would be
> subject to easier detection, while sound recordings (excepting ones
> that are done in special studios for the production of commercial CDs)
> contain in general always some ambient noise, which may render the
> decision between presence of stego or non-presence rather difficult.

I learned however that computer arts, in that no such comparisons
can be done, each piece being originally created by the artist/user
with (rather arbitrary) creativity, could apparently offer quite
some advantages as carriers of stego, since they could easily adapt
to external constraints, if required. [1]

M. K. Shen

-----------------------------------------------------------
[1] Some titles on computer arts:
B. Wands, Art of the ditital age. Thomas & Hudson, 2003.
C. Mason, Computer in the art room. JJG Publishing, 2008.
K. Guminski, Kunst am Computer. Reimer Verlag, 2002. (Thesis, LMU)
C. Klï¿œtsch, Computergraphik. Springer, 2007.

[Spinner]

Mok-Kong Shen <mok-ko...@t-online.de> wrote:

>Mok-Kong Shen wrote:
>[snip]
>> It seems to me that stego with sound files may have advantages over
>> stego with image files. For with the modern camera techniques,
>> pictures have barely random defects and so modifications would be
>> subject to easier detection, while sound recordings (excepting ones
>> that are done in special studios for the production of commercial CDs)
>> contain in general always some ambient noise, which may render the
>> decision between presence of stego or non-presence rather difficult.
>
>I learned however that computer arts, in that no such comparisons
>can be done, each piece being originally created by the artist/user
>with (rather arbitrary) creativity, could apparently offer quite
>some advantages as carriers of stego, since they could easily adapt
>to external constraints, if required. [1]
>
>M. K. Shen
>

That's not even english. The words are, but nothing else is. I'm
pretty sure our M. Shen is trolling again.
--
2+2!=5 even for extremely large values of 2

[Mok-Kong Shen]

Spinner wrote:

> That's not even english.

I suppose in a group that is not dedicated to discussions on languages
one could be a little bit tolerant about foreigners' poor English (if
one notes that 'english' is also not English).

M. K. Shen

[Spinner]

Mok-Kong Shen <mok-ko...@t-online.de> wrote:

Then how about 'not grammatically constructed as to make sense in any
known language'. However you translate it, the nouns and verbs do not
make a sentence that conveys information.

Typical nit pick response that's the fingerprint of a genuine troll.
It's always nice to decoloak one now and then.

[Mok-Kong Shen]

Spinner wroe:

> Mok-Kong Shen wrote:
>
>> Spinner wrote:
>>
>>> That's not even english.
>> I suppose in a group that is not dedicated to discussions on languages
>> one could be a little bit tolerant about foreigners' poor English (if
>> one notes that 'english' is also not English).

> Then how about 'not grammatically constructed as to make sense in any

> known language'. However you translate it, the nouns and verbs do not
> make a sentence that conveys information.

I can admit that I have the not very nice habit of writing unusually
long and complex sentences, often with embeded parentheses etc., and
that in the post in question I perhaps should have used 'since in them'
in place of 'in that' and maybe also put in some additional punctuation
marks to ensure unambigious parsing, but I think the meaning of the
sentence should nonetheless be clear to a normal reader. Now how about
this:

I learned however that computer arts (since in them no such comparisons

can be done, each piece being originally created by the artist/user

with (rather arbitrary) creativity) could apparently offer quite

some advantages as carriers of stego, since they could easily adapt
to external constraints, if required. [1]

Could you make sense out of it or nothing at all? If yes, please
re-read the original and give your critque to that once again with
some details.

M. K. Shen

[Noob]

Spinner wrote:

> Typical nit pick response that's the fingerprint of a genuine troll.

> It's always nice to decloak one now and then.

Plonk him, like some (most?) have already done.

> --
> 2+2!=5 even for extremely large values of 2

The sig delimiter is DASH DASH SPACE NEWLINE.

Yours is missing the SPACE ;-)

[WTShaw]

A programmer who writes a program that has ranges of options has
allowed those options by his selection. Who is the composer therein?
Such a production defines a new class of creation, rather a response
to a ripple started by the creator. Such constraints on the creation
could be used for stego purposes but I reserve the right to find some
of the intermediate products worth listening to in themselves.

In search of some stego capabilities years ago, I toyed with colors
and other image parameters which I could force on pictures. I found
worthwhile rewards in science and in art thus produced but I reserve
the right to pick what art I like and not promulgate that which I do
not find appeasing. The art side drew me into rather unexpected
circles and rather devoid of squares. (Pardont the idiomatic content
here..)

[mosherubin]

Hi M. K. Shen,

After reading the last few replies to your posting, I feel compelled
to comment.

I am amazed anew how abrasive, unpleasant, nasty, offensive,
loathsome, toxic, and petty some people can be. Truth be told, it is
the hallmark of a small, jealous, and angry mind. Even if one finds
the wording a bit hard to understand, there is a civil and cultured
way to ask for clarification.

Forums like sci.crypt are meant to stimulate a free-flowing stream of
ideas. The fact that there are those who try to poison the
atmosphere, have nothing constructive to contribute, and attempt to
stifle creativity (for reasons known only to them) will someday make a
fascinating psychoanalytic study. Just ignore the noise.

Your postings to date have always been thought-provoking and creative,
and you should continue posting for many years to come. I only wish I
knew a foreign language as well as you know English.

Moshe

P.S. To all flamers in the present and the future, save yourselves
the trouble and don't bother following up on this one -- it is
intended primarily for M. K. Shen, and is only being "CC'd" to
everyone else.

[Mok-Kong Shen]

It may be of interest to note that there is a very powerful graphical
programming language named Pure Data for the creation of interactive
computer music and multimedia works. See:

http://en.wikipedia.org/wiki/Miller_Puckette

http://crca.ucsd.edu/~msp/software.html

http://www.pd-tutorial.com

M. K. Shen