ZX Spectrum code on The IT Crowd DVD ?

[Dandu]

Hi

i have the DVD from the season 1 of The IT crowd, and the intro is a loading screen "Spectrum like" (i don't know if this is a real screen or not).

The video is here :

And there is a sound. I have tried to decode the sound, but with no success. I have tried with many software.

I have dumped the sound in WAV : down.dandu.be/ep00.wav

I'm not an expert, but i think it's a ZX Spectrum sound, because i have dumped many sound from vinyl for my blog (the page is in french : https://www.journaldulapin.com/2021/08/ ... e-vinyles/ ) and i like old computer code hidden into vinyl, CD, etc.

Can you help me ? Is this a ZX Spectrum sound ?

[Alessandro]

It is a Spectrum standard ROM loading sound in fact, and to me it sounds pretty much like part of a screen memory dump while loading, the pixel data that usually goes between addresses 16384 and 22528. The problem is, being just a small part, without header and the rest, you won't be able to load it into a Spectrum either real or emulated.

[zxbruno]

It is possible, as discussed in this topic (where I'm still hoping to get some help):

viewtopic.php?t=6978&hilit=code&start=10

I don't know assembler, but I'm sure someone could modify the following code to load data in the screen area as bytes instead of characters. The routine doesn't care if the audio has header or not. It displays whatever it listens too. Even if we ended up with a garbled mess, we could play with the width afterwards until we find something recognizable. It's what we do with sprite finding tools.

Code: Select all

        10         ORG 40000
        20         DI
        30         LD HL,#53F
        40         PUSH HL
        50         LD DE,#4000
        60 LBYTE   LD A,#7F
        70         IN A,(#FE)
        80         RRA
        90         RET NC
       100 NBYTE   LD L,1
       110         LD BC,#B201
       120 LBITS   CALL #5E3
       130         JR NC,LBYTE
       140         LD A,#CB
       150         CP B
       160         RL L
       170         LD B,#B0
       180         JR NC,LBITS
       190         BIT 7,L
       200         JR NZ,LBITS
       210         LD A,L
       220         CP " " <SPACE
       230         JR NC,OK
       240         LD L," " <SPACE
       250 OK      LD H,0
       260         ADD HL,HL
       270         ADD HL,HL
       280         ADD HL,HL
       290         LD BC,15360
       300         ADD HL,BC
       310         LD C,D
       320         LD B,8
       330 PCHAR   LD A,(HL)
       340         LD (DE),A
       350         INC HL
       360         INC D
       370         DJNZ PCHAR
       380         LD D,C
       390        INC E
       400         JR NZ,NBYTE
       410         LD A,D
       420         CP #48
       430         LD D,#50
       440         JR Z,NBYTE
       450         LD D,#40
       460         JR NC,NBYTE
       470         LD D,#48
       480         JR NBYTE

[flatduckrecords]

Here's my first attempt with @zxbruno's code. (The atricle in Microhobby is really interesting, but I haven't finished translating/reading it yet!)



Looks like something…might be there? Perhaps a graphics-finder tool would reveal more as you suggested @zxbruno.

(It's interesting how it responds to the little snippet of theme tune!)

And here's a control case, using the same code to load from a game tape, just to show it "working":



And here's the modified code (mostly the same, but with the character printing and ASCII-checking taken out)

Code: Select all

	ORG 40000

	DI
	LD HL,#53F
	PUSH HL
	LD DE,#4000	; DE=destination (in this case, screen display file)
LBYTE   LD A,#7F
	IN A,(#FE)	;Read keyboard & mic port $7FFE (bit 6 is mic, bit 0 is SPACE key)
	RRA		;The Carry becomes the bit leaving on the right
	RET NC		;ret if break pressed
NBYTE   LD L,1
	LD BC,#B201	;timing constant etc for LD-EDGE-2
LBITS   CALL #5E3	;LD-EDGE-2
	JR NC,LBYTE	;Cflag=0 if error
	LD A,#CB
	CP B
	RL L
	LD B,#B0
	JR NC,LBITS
	LD A,L
	LD (DE),A
	INC DE
	JR NBYTE

[R-Tape]

Nice work! I don't suppose you saved that screen? It might not be graphics so it'd be nice to see the bytes.

[flatduckrecords]

Nice work! I don't suppose you saved that screen? It might not be graphics so it'd be nice to see the bytes.

Nothing jumps out to me in the hex view but I’ll upload the file shortly in case anyone wants to take a look.

[flatduckrecords]

Files:
titcdvd.scr
rough disassembly

edit - link fixed. (Thanks Mark and JP)

[1024MAK]

The “rough disassembly” link just gets me a 404 error

Mark

[jpnz]

The “rough disassembly” link just gets me a 404 error

Mark

https://robertmorrison.me/public/files/titcdvd.scr.txt

[zxbruno]

Very nice!

I imported the .scr file to ZX Spin and played with the height and width, but so far no luck.

[Dandu]

Thank you for your help !

[R-Tape]

It doesn't make any sense as code, but some of it looks like it's trying to be graphics. I can't take a pic just now, but viewing it in 8 pixel chunks there is structure and pattern from the off. Could still be gobbledegook of course.

Isn't it more likely to be actual Speccy loading noise rather than something entirely fabricated? I would have thought doing the latter would be harder.

[1024MAK]

My thoughts…

It doesn’t look like a machine code program to me. Even ignoring the start, after a run of NOPs it makes no sense. And it does not look like pseudo random data.

It could be graphic data, but at which position it starts in screen memory, if it is screen data, is another matter.

Mark

[Sokurah]

I can't take a pic just now, but viewing it in 8 pixel chunks there is structure and pattern from the off. Could still be gobbledegook of course.

It could be leveldata, which could give it a repeatable structure, but be much harder to identify than code or graphics.

[MatGubbins]

How do we bring up pages of loading screens on Spectrum Computing?

[R-Tape]

How do we bring up pages of loading screens on Spectrum Computing?

In the search, under 'results layout' you can select 'load screens only'. Is that what you mean?

[zxbruno]

If anyone has most of the existing Spectrum software collection on their hard drive, would it be possible to look for this block on code on that hard drive? Someone a few months ago said it would be easy to do if we were looking for part of a loading screen.

[Mpk]

We should have a contest to guess where it's from. My money's on Stonkers.

[flatduckrecords]

We should have a contest to guess where it's from. My money's on Stonkers.

If it turns out to be another Pringles advert I don't know if I'll be upset, or impressed!

[Van_Dammesque]

The .scr and post by Bruno show a few lines that should not be there due to the music playing, can you upload a scr without the music effect, i.e at 24 seconds of your video?

[+3code]

The .scr and post by Bruno show a few lines that should not be there due to the music playing, can you upload a scr without the music effect, i.e at 24 seconds of your video?

Agree, the last 8 or 9 tones at the end of the video seems beeper music. They make the punctuated lines.

[MatGubbins]

This little program will copy the data to the screen at every location (without the crappy music bytes at the end) and eventually shuffle into the attributes and crash. Pressing break will exit to basic, run again and it will start again.

Is there anything there within the data? Possible, but it seems just to be data for a game, levels or partial screen data combined with rubbish.

https://www.sendspace.com/file/3fjbx5

Have fun with IT (Crowd).....

[flatduckrecords]

The .scr and post by Bruno show a few lines that should not be there due to the music playing, can you upload a scr without the music effect, i.e at 24 seconds of your video?

Yeah I actually went back to @Dandu's extracted audio to try again afresh (with the music clipped off as well). I had been using a tape recording from YouTube so I tried directly playing the .wav from computer into the Spectrum (and tried a couple of different Spectrums for good measure!)

Re-ran it a few times but I get different results each time. I guess because there's no leader tone, it's not possible to say where the byte boundaries are within the bit stream? Even loading the .wav directly into Fuse generates different bytes each time.

Any errors are ignored and those erroneous bits will be dropped—which would mean the following bits would then 'shuffle up' and be offset from where they should be. That would skew the bitmap part of a SCREEN$, but I think it would totally scramble any machine code.

Anyway here's the .tap file I've been using if anyone wants to try (thanks for posting yours @MatGubbins). (And sorry, I should have posted that sooner!) If you use a DivMMC: once the program has loaded remember to BREAK in and issue .tapein -c to close the tape file and allow loading from the Ear socket.

byteloader.tap

48k_1
48k_2
Harlequin1
Harlequin2
Fuse1
Fuse2

I think the Fuse results look the most structured, but it could just be some sort of UDG pareidolia that makes me think so.

[Sokurah]

Too bad that Graham Linehan was banned from Twitter - we could've asked him if he knew.
Does anyone know if he's on Instagram or Facebook?

[MatGubbins]

So you're getting different data bytes each time and a different byte count too! Oh crap!

Here's all 6 of the files processed in the same style as before. They will crash when the attribute area starts to fill up.

https://www.sendspace.com/file/xglcpb

I've looked for sprites within the data, but nothing stands out either.

Maybe it just needs that one hit at getting the correct data sync and things might pop into line. Some of the data does seem to form a title logo/shape when the scrolling is paused at a certain time, but eyes and the brain do play those tricks of filling in the blank spaces.