Harrier Attack / Converting from poor tape

[wesmond]

Hi all,

 

Regarding this game - Harrier Attack by Durell Software:
http://www.atarimania.com/game-atari-400-800-xl-xe-harrier-attack_37885.html

I can confirm that "something Atari" exists for it, and I got that "something" from ebay. The inlay looks like it was originally an Oric inlay, with an Atari sticker put on top, which has mostly worn off. There are no instructions whatsoever - blank on the inside. Quality of the tape is poor. Evidence of some other machine type before/after the Atari recording. To my surprise though, it loaded on a real 800xl/1010 - start+option, first stage has a couple of reasonably crafted pictures of harriers, and it loads in about 5 minutes. The game itself plays a beep, lets you type your name into some kind of table, and then starts to horizontally scroll a blocky aircraft carrier before (possibly) crashing. (Or perhaps I just don't know what to do with no instructions - keys/joystick did nothing...).

 

I'll get some scans/photos up in due course, but wondered if anyone's got any hints for getting really terrible tapes converted. I have tried half a dozen Audacity recordings from a walkman - method usually works on half-decent types. I can get a reasonably good looking level without clipping, but the quality coming through is awful with swirly frequency sweeps going on. Wav2Cas says...

"initial sync header is too noisy, found unexpected bit transition..." - I'm not surprised, it's a tall order. But I guess there's hope if the 1010 can do it.

Alternatively, if I can load it on real hardware, is there software to get that into a CAS format from the Atari? I used to use Transdisk IV in ages past, although I think that produced a non-standard OBJ file needing the Transdisk loader. So something more accessible than that would be good...

 

Edited April 24, 2021 by wesmond
typo

[www.atarimania.com]

Holy crap!

 

So it does exist?

[+x=usr(1536)]

Impressive find!

 

47 minutes ago, wesmond said:

I'll get some scans/photos up in due course, but wondered if anyone's got any hints for getting really terrible tapes converted. I have tried half a dozen Audacity recordings from a walkman - method usually works on half-decent types. I can get a reasonably good looking level without clipping, but the quality coming through is awful with swirly frequency sweeps going on.

The audio quality may be down to the Walkman in question.  Try the usual stuff: cleaning, aligning, and demagnetising the heads and capstans, checking that any belts are still OK, use a different cassette deck (if possible), etc.  Also make sure that any type of noise reduction such as Dolby is off, and that the tape type is not set to chrome.

 

Some post-processing may be able to reduce the frequency sweeps, though they likely can't be removed entirely.

 

If the 1010 was able to get a successful read, you could try tapping the audio output from it during a successful load and seeing if that nets a better .wav file.  Also, given that the 1010 was able to load the cassette successfully, my best guess is that something in the Walkman isn't giving a usable capture.  Give the Walkman a good going-over and see what happens.

[Sugarland]

Thank you for trying to recover it.

[+x=usr(1536)]

A couple of other thoughts crossed my mind:

• Are you capturing the cassette as a mono or stereo .wav file, and at what sampling rate?  The right channel is typically used for data, with the left channel reserved for audio.  If you capture only the right channel, that may help to remove any noise found in the left channel, which would lead to a better-quality capture - but it is also going to be dependent on the Walkman having the heads properly-aligned.  If they aren't, noise in the left channel could be polluting the data stream in the right channel during playback, leading to a bad capture and failed load.
• To tap the audio from the 1010, it may be possible to open up one end of an SIO cable, separate out the wires for pins 5 (data output) and 11 (audio output), connect the other end of the cable up to the unused SIO port on the 1010, and record from there.  Since SIO is completely pass-through, I'd expect the audio to be mirrored on the port that isn't in use.  No idea if this will work or not as I've never tried it, but it should be possible in theory, at least (which is where someone tells me that this is a terrible idea and why I'm completely wrong )

In any event, good luck - and whatever you do, don't give in to the temptation to bake or freeze that tape no matter what anyone may say.  That's seriously last-resort stuff, and given that the 1010 was able to read it, you likely don't need to do anything that drastic with it.

Edited April 24, 2021 by x=usr(1536)
Correct output pin numbers FTW

[wesmond]

Thanks for the hints - will give it a try. The walkman certainly doesn't get much use these days, although to be fair, neither does by 1010...! If you POKE 54018,52 and then record audio straight out of the Atari, does that work?

More news... IT'S ACTUALLY PLAYABLE!!! You start by typing 1 (return), then your name (return), and a level 1-5 (return...) and after that, it's console keys. You push Option to sort of go up, Select to sort of not-go-up, and Start to shoot some kind of projectile that shoots things in the air for a while, and then drops to the ground and blows things up there as well... the graphics remind me of Airwolf - sort of blobby green hill shapes.

Photos attached - sorry about the quality. Photoing the TV with one hand while flying...! I'll scan properly and post in the AtariMania forum when I'm done, if that's best way?

(A pleasure @Sugarland - I was excited to see it, and glad it's turned out to be something interesting. I feared it might all come to nothing...)

[djmat56]

If you are happy to post the cassette to me here in the UK I would be more than happy to try and dump it and convert. I have a good setup and have already contributed around 100 tapes to the community

[wesmond]

Here's a link to my best WAV so far...

https://we.tl/t-5H1pCJaFB0

I'm capturing stereo 44100, both channels seem to be the data.

[wesmond]

@djmat56 - thanks - I'll try the suggestions above, and will come back to you if I'm defeated!

The right-channel is a good idea...  Don't worry, I won't freeze/bake/tumble dry...  

My walkman is a Sony TCM-313 - it has no clever buttons basically! Unfortunately that's the only tape deck I have besides the 1010.
 

[www.atarimania.com]

This is quite incredible... However crappy the game may be, I'm amazed this is the first copy ever found in almost forty years. I mean, the program sold tons of copies on other platforms so is far from unknown. Maybe it was just test marketed on the Atari with a very small run and then dropped due to poor sales or pre-orders?  

[_The Doctor__]

Many vendors didn't give stuff enough time for word of mouth to get the ball rolling... by the time a person finally heard of a thing... it was already done for. Sort of like many of the products people do on AA today with short 1 or two week windows to get in on it... it can take months before folks have the time to get to their hobby in this day and age... when they see all they missed they get bummed, lose their joy and it could be months before they stop back in again.

 

such is life. now imagine when all you had was a note here or there in some magazine that most people didn't buy/get/receive.....

 

 

[wesmond]

Certainly seems they didn't advertise it very much! The other games in the ebay box are pretty common (https://www.ebay.co.uk/itm/254947472476). Would be fun to know more of the story behind it. For now, just glad to have a chance to preserve it. ?

[www.atarimania.com]

A very big thank you, this is quite a find 

[manterola]

I will also try to produce a. CAS from the wave file. 

[www.atarimania.com]

Pretty cool that there's a loading screen as well (taken from the Spectrum version it seems).

 

I tried looking up for the programmer who converted the game (R. Merson) but came up with nothing...

[+x=usr(1536)]

1 hour ago, wesmond said:

If you POKE 54018,52 and then record audio straight out of the Atari, does that work?

Maybe. I'd still recommend tapping the SIO pins I referenced earlier rather than holding a mic up to the speaker, using audio from the monitor output, etc.  Basically, you want to capture as close to the 1010's output as possible - anywhere downstream of that may introduce audio distortion, etc. that would leave you with an unusable capture.  In-between the 1010 and the 800XL's SIO port would be ideal, but the breakout cable I mentioned earlier may be a possibility as well.

 

Conversely, there are other points within the 1010 itself that could be used to capture the audio.  But given that we know that the 1010's output is acceptable to the 800XL, in this case it's probably best to take it from SIO.

1 hour ago, wesmond said:

Here's a link to my best WAV so far...

https://we.tl/t-5H1pCJaFB0

I'm capturing stereo 44100, both channels seem to be the data.

I'll convert it to .cas and see what I can do with it, if anything.  Don't have any blank cassettes to hand, but I can use the FujiNet to work around that.  Just something to keep in mind as I may end up with a version that works perfectly fed from the FujiNet but that can't be used to re-create an actual cassette.

 

BTW: since both channels look like they contain data, try capturing each one individually - i.e., if capturing the right channel, don't capture the left channel and vice-versa.  It's possible that crosstalk between the two creates a summed waveform that wav2cas doesn't like, but one or the other channel taken individually might be usable.

Edited April 24, 2021 by x=usr(1536)
Moar infos

[nysavant]

Wow.  I had no idea Harrier Attack ever existed on the A8, what an amazing find!

[wesmond]

Thanks to all who are trying out the WAV file!

I'll try phono out from the 800xl first, as I think I can do that fairly easily if I dig out a couple of cables... The other method I thought of this evening, was to write some code on the Atari to read the raw data for each of the two stages into RAM, and dump the bytes into a couple of disk files on SIO2SD. Then fire up Altirra, read those disk files back in and dump them out to the (virtual) cassette in authentic real-time mode... and that should produce a perfect WAV of the original. I don't think there is anything sophisticated going on in the loader/code, so hopefully just a case of remembering/learning how to do block-reads... (If anyone has some code/pointers to some right at their fingertips to do this, that'd be great - but otherwise I think I can do it)

[+x=usr(1536)]

Spent some time poking around in this, and noticed a couple of things.  For comparison purposes, I'm using the Harrier Attack .wav file linked to in post #8 of this thread, and one generated from a known-good .cas file - in this case, Chuckie Egg.cas.  They're both attached below for reference.

 

Starting off, here's what Chuckie Egg looks like:

Note that both the waveform (top) and spectrum analysis (bottom) exist within tightly-defined boundaries and appear as solid streams of sound.  Granted, this is due largely to how cas2wav generates .wav files from .cas images - it'll basically spit them out as ideal representations to feed into the Atari, not necessarily as they were actually captured.  However, they're useful for getting an idea of where things should be falling in terms of .wav captures.

 

With that in mind, let's take a look at the Harrier Attack waveform and spectrum analysis.

 

I'm going to guess that this is unprocessed audio, basically as it was when captured.  Notable items:

• It's stereo, not mono
• The spectrum analysis is showing what I can only describe as reflections of the audio prior to the 2m17s mark in the lower part of the band
• The audio is very hot in places - from watching levels on the dB meter, Chuckie Egg replays at a steady -8dB, but Harrier Attack peaks well above that at times (0db or slightly above)

Unfortunately, this all points to a capture that isn't currently usable without reconstructive surgery being done on it.

 

My suggestion: recapture the audio.  If possible, do two separate captures, one of each track indivdually.  If that's not doable, capture in stereo but separate the tracks out into two mono tracks in post.  And the volume still needs to come down.

 

Don't try to recreate exactly what is seen in the Chuckie Egg screenshot.  Better is the enemy of good enough in this case, so just use that one for reference in capturing.  If you can get your waveform and spectrum analysis within the same boundaries (including level) that Chuckie Egg uses, you'll have a better shot at converting it to .cas later.

harrier_attack.wav Chuckie Egg.wav

[+x=usr(1536)]

14 minutes ago, wesmond said:

I'll try phono out from the 800xl first, as I think I can do that fairly easily if I dig out a couple of cables...

Can't hurt to try.  I'm still of the opinion that capturing directly from the 1010 would be preferable, but also understand the practicalities attached to that.  My one suggestion would be to make sure that the phono out from the 800XL has clean audio (e.g., no pops, hums, cracks, etc.) before committing anything to disk.

15 minutes ago, wesmond said:

The other method I thought of this evening, was to write some code on the Atari to read the raw data for each of the two stages into RAM, and dump the bytes into a couple of disk files on SIO2SD.

I may be misunderstanding, but it sounds as though you're talking about transferring the contents of RAM to a disk image.  While that will definitely get the digital data committed to media other than cassette, it's only likely to be the runtime code and data for the game.  For a valid cassette image, you'd also need to capture the loader, etc.

18 minutes ago, wesmond said:

Then fire up Altirra, read those disk files back in and dump them out to the (virtual) cassette in authentic real-time mode... and that should produce a perfect WAV of the original.

Based on what was mentioned above, I don't think that will work.  The virtual cassette device will (I imagine) be expecting to see a loader, etc., not the contents of what would effectively be a memory dump.

 

Having said that, from a preservation standpoint, I am by no means opposed to having the actual data sitting on other media as a safeguard; that is absolutely better than having it on just the one known tape.  But if the goal is to create a digital image of an analogue tape that could be used to generate a new cassette capable of being booted on original hardware, this approach is not going to be able to do that.

 

Either way, there are

[wesmond]

I'm not suggesting doing a ram freeze of the running code. I'm suggesting loading the raw cassette bytes into RAM with a little IOCB block read code, writing those bytes into a binary disk file just so I can take that to a PC running Altirra, then more code to block-write those bytes back to virtual cassette. We should end up with a "perfect" version of what the 1010 is able to read. I am impressed with the 1010 - I tried recording from the phono, and the quality was cleaner than the walkman for sure... it reproduced with even greater accuracy all the horrible stuff on the tape! Somehow the 1010 is able to load it...
 

Edited April 24, 2021 by wesmond
typo

[+MrFish]

Can't you just use one of the cassette to disk copying programs?

 

I haven't used them myself, but here's one I found in the Holmes archive.

 

Cassette to disk copy.atr

 

 

[firebottle]

Knowing that the 1010 reads it OK, a quick listen to bits and pieces of the recording reveals some wow & flutter... I'd bet the Walkman needs a new belt.

[+CharlieChaplin]

46 minutes ago, MrFish said:

Can't you just use one of the cassette to disk copying programs?

 

I haven't used them myself, but here's one I found in the Holmes archive.

 

Cassette to disk copy.atr 2.02 kB · 1 download

 

 

 

 

Well,

 

I do not know why, but Dean Wittman's program never worked for me and I could not copy a single program with it onto diskette. In the meanwhile I have tested, recorded onto CD-RW and tried to copy 821 tapes. For a simple tape to disk copy I have mostly used Copy C/D 54k by M.Thiel which works on 64k XL/XE computers under DOS 2.5. This program will only copy tapes with standard block lengths (128 / 132 Bytes) and does not work with the non-standard ones by Novagen, English Software and others...

 

The attached disk image contains CD-Copy, boot it with or without Basic, when the green menu appears move the cursor with a joystick or keyboard (cursor-keys), select CDCOPY.COM and press START or fire button to load it. Insert a new disk (if you forgot to format it, CDCOPY can do this for you) and press A to read the data from tape and later D to save the data onto diskette. 

 

 

CDCOPY.zip

[wesmond]

Easier to write my own for something simple like this. I've got the bin files done - just some faffing around with SD cards and I should have something later today...

 

Ths issue is that we don't want cassette to disk, we want cassette to... nicer cassette, preserving the multistage experience...

Edited April 25, 2021 by wesmond
More detail