Possible 2nd Layer of Sounds

From Portal Puzzle Wiki

Here is another lead to follow. It would make sense that some of the sound files have two separate signals, especially with the voices and other things heard on some of the files. Link goes to the original post. Portal Secondary Signals

Contents 1 The Speculation 2 Pink & Blue 3 A Closer Look 4 Dinosaur 20 5 Dinosaur 20: HPF 6 Dinosaur 20: FFT 7 Work To Be Done / Speculation 8 8 bit Left/Right Channel Split on Dinosaur4 9 Channel Split on dinosaur_noise 10 Misc. Stuff 11 Thanks

[edit] The Speculation

I left for work this morning hoping this would be solved by the time get I get off work, but apparently the original thread's getting buried amidst all the hubub. It seems like the major summaries have not picked up on this thus far, so I'm reposting it to get more attention and technical help.

I wish to make clear that this is NOT JUST the SSTV images, but IN ADDITION to the image data. Like I said in the original thread, SSTV signal is highly resistant to noise, which makes it great for embedding additional data.

I found a second (possibly analog audio) signal multiplexed with the SSTV stream.

[edit] Pink & Blue

The pink and blue are inherent in the file, I did not alter this picture aside from selecting spectrum display in audacity. The pink areas are when the SSTV signal is active. The blue area at the end is where the SSTV signal has stopped, and you can see the second signal I was talking about (the pink wave form).

[edit] A Closer Look

I drew some boxes in case that was a bit hard to see. The blue box across the bottom encompasses the primary SSTV signal. The black box on top encompasses the second (perhaps analog) signal. I also tried to trace some of the second signal within the pink area in black. The black arrow points to the second signal within the final seconds (blue area) of the file.

[edit] Dinosaur 20

Here is a zoomed in section in dinosaur20 from 17s-20s.

[edit] Dinosaur 20: HPF

Here is the same 17-20s segment of dinosaur20 with a simple high pass filter (2k). It provides a bit better contrast for the second signal on top.

[edit] Dinosaur 20: FFT

Dinosaur20 FFT "higher quality FFT of dinosaur20, from FL Studio's Edison editor." - by akx

• Note: Page F's up when I tell the Wiki to create a thumb of the fft.png or fft.jpg file.

[edit] Work To Be Done / Speculation

I'm hoping someone with the appropriate sound or data expertise would step forward and demultiplex these signals into their component streams. I have no such expertise, only a somewhat sharp ear. Please, I really want to know what it is!

Some suspicions why this might be key to something:

• "Changed radio transmission frequency to comply with federal and state spectrum management regulations" patch note might actually be a hint to bandfilter/demultiplex the sound files before converting them to SSTV images and some yet unknown data.

• The infamous "noise" files sound like they have a complex harmonic. Perhaps the combination harmonic of the secondary signals will neutralize the noise and make the noise files intelligible.

• Video Source (deep-link to most relevant section of video http://www.youtube.com/watch?v=nM9QoP3ghlo#t=1m44s)
  

• Perhaps the second signal, when appropriately sped up or manipulated, pronounces a single alphabet. This might give meaning to the seemingly random insertion of morse code in the original sequence of 26 sound files. [1]-2-3-4-[5]-6-7-8-9-10-11-[12]-13-14-15-16-[17]-18-19-20-21-22-23-24-25-26

[3 letter word] [6 letter word] [4 letter word] [9 letter word]

• Edit by Looploony1 on Youtube*

I think Portal 2 and HL2:Epi3/HL3 are going to be taking place at the same time and the two worlds are being "Combined" (little pun there) anyway see video above to see a short moment when Portal and Half-Life meet (happens in HL2:Epi 2).

[edit] 8 bit Left/Right Channel Split on Dinosaur4

I did some basic hex manipulation on the dinosaur4 wav file:

• increased the number of channels from 1 to 2
• lowered the bits per sample from 16 to 8

Here is a header diagram I made in case anyone wants to duplicate this:
http://www.mr-green.nl/results/dinosaur4_header.jpg

mono 16 bit dinosaur4 (original):
http://www.mr-green.nl/results/dinosaur4%20(original).wav

stereo 8 bit dinosaur4 both channels:
http://www.mr-green.nl/results/dinosaur4%20(double%20channel,%20half%20bits%20per%20sample).wav

stereo 8 bit dinosaur4 right channel:
http://www.mr-green.nl/results/dinosaur4%20(R%20channel)%20sstv.wav

• This remains a viable SSTV signal, which scans into this image:
  

http://www.mr-green.nl/results/dinosaur4%20(R%20channel)%20sstv.png

stereo 8 bit dinosaur4 left channel:
http://www.mr-green.nl/results/dinosaur4%20(L%20channel)%20modem%20carrier.wav

• This is nothing like a SSTV signal, which makes me think we might be onto something. In fact it kind of sounds like a modem carrier.
• As I pointed out in IRC, this sounds a lot like a program encoded on an audio tape by a Commodore 64 or similar system. -Karride

Try to run the fizzle noises through that 16-bit converter thingy. -Thepwnedgod

[edit] Channel Split on dinosaur_noise

More proof for this theory: dinosaur_noise.wav contains two different (but consistent) noises when split in this way. Look at the waveform for http://files.dotbas.net/dino/dinosaur_noise_stereo.wav in Audacity, and compare to the waveform for the normal dinosaur_noise.

In addition to this, if you zoom in on the waveform of the file downloadable above you can see quite clearly that the right hand channel is a binary square wave waveform, the problem comes in decoding it... - GITs

It appears that all negative sample values have been moved up, and all positive values have been moved down. It should be possible to revert that process.

Having run this throuh a C64 emulator I have managed to get nothing out the other end... - GITs

It actually looks like a 2's complement overflow. It is possible that the sound was amplified without saturation mode, resulting in all values changing to the opposite polarity. It is the result of changing from 16bit mono to 8bit stereo. - RC-1290

[edit] Misc. Stuff

A Sample of pure SSTV signal stream.

Link to all the dinosaur wav/image files.

Link to a SSTV software to decode the wav files.

link to tool to encode text into wav files for TTY machines

link to tool to read text out of TTY recorded sounds

[edit] Thanks

Huge thanks to Reddit user 'pokute' for the possible lead. Let's see where this goes guys!

The end of a few of the dinosaurnn.wav sounds just sound to me like the typical interference you get when listening to a radio amateur receiver. I wonder if they just transmitted the SSTV images to, effectively add the typical noise you'd get and make them look more authentic? If so it's quite likely there is more than one signal, but unlikely that it's anything to do with this update. (OTOH, they may just have mixed in some noise to do get the same effect)

If you think about it, with hindsight, nothing so far has been less than obvious to work out. Morse code - that's easily decoded. SSTV, easily decoded with a plethora of available software. The MD5 of The quick brown fox? google tells you that in seconds. The hardest part, perhaps, translating the MD5 into the phone number, although there were some pretty obvious clues the noise in the SSTV images didn't help.

The, again with hindsight, user backup password backup was stated in an obvious way, albeit it took some time for the penny to drop (probably had more to do with folk thinking too hard rather than seeing what was given)

Once that was obtained the ascii art is easily obtained.

So the idea they have some deep mysterious hidden signal to work out doesn't fit. Folk are thinking too deep now. Everything so far has been obvious and relatively straightforward to interpret and obtain. GCHQ wouldn't break a sweat with this :)