Use this forum to submit new files for the download section of the archive. I will check each submission and upload it to the archive on a regular basis.
I've attached an updated version of Dragon Convert for Windows.
This update (v0.2) adds support for loading WAV files and saving CAS files. I've also introduced some improved options for handling leaders from WAV files, so that the user can choose whether to fix them at a specific length, or keep the original (variable) lengths.
I still haven't provided any help or instructions, but hopefully it's not difficult to use...!
Note: You'll need Microsoft's .NET framework (version 4) installed to run this release of DCWIN; there's also a build for .NET 3.5 attached further down the page!
I'd be grateful for feedback on DCWIN please, if anyone's given it a trial (especially if someone's had success with it)!
The only testing I was able to do was with WAV files that had been back-converted from CAS files, although hopefully it should work at least as well as the old MS-DOS version...
No worries, thanks anyway. If it helps, DCWIN doesn't need the full framework package, just the minimal "Client Profile" release is sufficient.
New installations of Windows 7 can already have .NET 4 included, whilst other Windows 7 systems should at least have .NET 3.5 (as required by the previous version of DCWIN, however I built this new one with Visual Studio 2010 running on Windows XP, which only targets .NET 4 unless the older frameworks are added manually). Incidentally, it's also possible to install and run everything inside Microsoft Virtual PC without any problems... that's where my development environment currently resides!
I've attached a release built for .NET 3.5, hopefully good if anyone's running Windows 7 yet reluctant to install the .NET 4 framework!
I've been trying your program and I've found the following issues:
- With real tapes, which are often slightly damaged, it doesn't get much useful data out of it.
- With tapes produced by my program, it fails to calculate volume or wave length. Maybe you treat 8-bit samples as signed, when according to the wav standard, 8-bit sample files are always unsigned and 16-bit samples are always signed.
- I find a bit confusing the WAV to CAS parameters volume and wave length, because volume can only be taken to 126 (for me, a sign that you treat 8 bit samples as signed), and also because the 0 waves and the 1 waves have different length. In my tests at 44100hz, the 0 waves are 40 bytes long and the 1 waves are 20 bytes long. So I don't really know what value should I put there.
- Altough it works very well, I've found that when you convert from CAS to WAV, the resulting file doesn't load in a real Dragon. I had the same problem while developing my own program and I got over it by adding 8 extra sync bytes (0x55) at the end.
On the bright side of life, I have to say the program is nice and easy to use.
I've attached the wav file of the Space Shuttle game as written from my program. It loads well under XRoar and on a real Dragon, so you can test it.
I'll be trying to recover a lot of games these days, so I'll upload you a real wav dump for you to test in the following days.
Thankyou Rob, that's helpful feedback! I appreciate your comments!
Hopefully DCWIN can cope with those damaged tapes at least as well as DC does, however I'm aware that they can struggle with such troublesome recordings. I've been considering including at least one alternative conversion algorithm in the next version, that would give users a choice of trying either the existing 'wave trace' method or a simpler positive/negative transition method. Maybe that'll help... or maybe not, I don't know!
I can confirm that DC & DCWIN both treat the samples as unsigned (and expect them to be 8-bit mono). I had a pretty unsuccessful quick go at converting the Shaper sample that was posted here recently, but didn't check whether it was the correct format.
The setting for "volume" refers to a threshold used to identify waves by indicating the minimum amplitude to expect waves to reach (it's therefore largely insignificant for WAV files that have been created by conversion from CAS files, since their waves are likely to be loud and well formed; your Space Shuttle file actually seems to convert perfectly with DCWIN if even a tiny fixed volume threshold value is chosen).
The "auto" option is new for DCWIN and works by quickly surveying all the data for wave peaks and troughs, averages their heights and finally takes a quarter of that result. Surprisingly, the reason that "auto" didn't work with your Space Shuttle file, turns out to be due to a curious property of that file together with a minor flaw in the design. This version of DCWIN is surveying the data by examining every 1/100 byte, evenly spaced, therefore it looks only at those bytes with offsets 44, 144, 244, 344, 444, etc during this pass, which as it happens are all either $7E or $7F in the Space Shuttle file, so it rather unluckily saw a completely silent subset of values throughout the entire length! Clearly, I should have used a prime number for the step size of this surveying process instead of 100; having retested with the number changed to either 97 or 101 it works just fine for the Space Shuttle file, so I'll fix this fault for the next release, in fact also switching to a somewhat smaller number for better accuracy!
The maximum value for the volume threshold is relative to the centre point of the waves, which it expects to be $80 and from this it attempts to trace values occupying the range $80 +/- (Vt + 1) where 'Vt' is the volume threshold setting. The setting for "wavelength" is simply their average, so 30 would be an ideal value for your waves of 20 and 40 samples each.
I'm puzzled by your comment that generated WAV files don't load back in a real Dragon without extra sync bytes being added at the end. I don't remember hearing of that problem before and can't immediately think of a good reason for it to happen whilst files convert back and forth between WAV and CAS without loss, but I can certainly provide an option to add trailing sync bytes if that helps users!
I did a single attempt at sampling Katerpillar 2 and converting it to a CAS file. First attempt failed, probably because I used 16 bit samples. Once the WAV file was properly converted to 8 bit mono samples, I got a clean CAS file that loads into XROAR. Result is attached.
