Altirra and ways to CSAVE

[morelenmir]

I appreciate there is no reason to save anything to 'tape' from inside Altirra and also that Phaeron is both strongly opposed to implementing the facility directly and dislikes cassette data recordings on a moral, philosophical and existential level. Fair enough.

 

HOWEVER. Assuming one was insane enough to be udaunted and STILL wanted to do so... I wondered if it was possible to fudge the process together using the 'Record Audio' functionality?

 

You CAN issue a 'CSAVE' command, save an audio recording to *.WAV and press return. You get a wave file that holds the correct lead-in whistle along with the usual, muffled program buzzes and creaks you hear when monitoring a data recording over television speakers on real hardware. As it stands, however, if you then try and load that *.WAV in tape control, cue up to the correct part of the recording and CLOAD - it doesn't work. Therefore I wondered if it was possible to massage the wave file in Audacity or SoundForge so the data 'sounds' are prominent enough to work?

 

In the stoneage past, SOME people - SWIM - I have it on good authority, I am told by others, my best friend's sister's cousin, etc; Ahhemmm... used to share their old ZX Spectrum tapes over the telephone in a similar way... Those copyright-disrespecting madmen!

Edited July 11, 2014 by morelenmir

[thorfdbg]

HOWEVER. Assuming one was insane enough to be udaunted and STILL wanted to do so... I wondered if it was possible to fudge the process together using the 'Record Audio' functionality?

No, this won't work. There is a different signal on audio-out than on serial-out. The reason is that pokey *also* uses the audio channel to drive the serial output line, i.e. ser-out only receives the mixture of the mark/space audio registers, whereas audio-out *additionally* recieves the signal from the pokey registers that drive the serial timer.

 

I tried this once with Atari++: You can save to a wav, and get the precise output, but the signal is not the same as the one you get from the tape recorder. Specifically, wav2cas does not work on this.

[Rybags]

Just tried muting the 2 "timer voices" - problem is that when the tape handler writes each record it reinitializes them.

Also, clearing the NOISYIO flag ($41) doesn't help, that gives no audio output at all.

 

What might work is a custom OS hack that sets the mark/space voices (1-2) as normal but mutes the timer voices (3-4)

[phaeron]

The clock signal isn't really a problem, at least at the standard 600 baud rate. It's far away from the data tones in frequency and is additively combined, so it's easy to filter out with a high pass filter. That's assuming that it's even noticed by the decoder at all. Modern decoders using bandpass filters instead of zero crossing detection will already filter it out.

 

The real problem is the phase discontinuities. The serial port output uses one divide by two for all output, whereas the audio output has an independent divide by two for each channel. The result is that the signal jumps randomly on transitions between 0 and 1 bits, introducing lots of noise. Current decoders will filter a lot of this out, but it's still pretty likely that a few bit errors will sneak though. I once received a "tape" recording that was generated off the audio output and managed to decode most of it, but a couple of the blocks were still marginal.

 

My instinct is that it's possible to make a decoder handle such a recording, but even if it were successful, it would have reduced error margins compared to a standard tape recording.

[morelenmir]

Obviously the Spectrum was a much simpler machine! SWIM had an 20-30 tapes of coped spectrum games, transmitted over the telephone from his uncle in Portsmouth! MOst of them even worked! I have heard,