Jump to content
  • entries
    322
  • comments
    878
  • views
    250,197

Are you ready for this? pt 2

Sign in to follow this  
EricBall

750 views

Second try at making the VCS play 2 Unlimited's "Get Ready For This" (or at least 30 seconds of it).

 

Differences:

1. First 10 seconds are skipped, which means it starts with the better sounding chords.

2. Only uses AUDC 4, 6 & 12 (square waves)

 

Notes: it sounds better on a VCS than an emulator (although the emulator does the score color, while my 7800 doesn't, weird); not compatible with an unmodified SuperCharger.

 

Method behind the madness:

1. Song ripped from CD to WAV, then band pass filtered to 60Hz-15kHz and resampled to 31.4kHz 8-bit signed mono.

2. Each 524 sample (one frame) chunk is then brute force convolved with TIA waveforms to determine the two best matches (ignoring phase, since it's virtually impossible to control the phase of the TIA audio waveform). The best matches are the saved to a binary file and assembled in with the player.

 

I'm both impressed and disappointed with how well this works. On one hand, the process does seem to work for simple chords - the more notes / instruments the more it seems to fail. But, there's also a lot of noise even with this simple test which I suspect can't be easily removed. And I guess it was too much to hope for speech to be recognizable.

Sign in to follow this  


6 Comments


Recommended Comments

You are changing AUDxx once per frame, correct? How much better would it sound to go to two or three times per frame?

Share this comment


Link to comment

You are changing AUDxx once per frame, correct? How much better would it sound to go to two or three times per frame?

It might, but I don't think it will get much better unless you instead bang on AUDF and do 4-bit PCM instead. (Of course, that chews up ROM really, really quick.)

 

Double the data rate, double the ROM required (or half the time). But that also doubles the lowest frequency which can be determined using my method*. I suspect this would also pick up more high-frequency false noise.

 

 

* Note for those interested, there are 123 NTSC TIA AUDC/AUDF combinations which have a fundamental frequency of less than 60Hz.

Share this comment


Link to comment

Hm... it sounds like putting a C64 tape into an audio cassette player. Until you said so in this message, I'd never even thought it's supposed to be music, let alone "Get Ready for This" ... :lol:

Share this comment


Link to comment

Hm... it sounds like putting a C64 tape into an audio cassette player. Until you said so in this message, I'd never even thought it's supposed to be music, let alone "Get Ready for This" ... :lol:

Yeah, my wife said the same thing. Once she knew what it was, she could hear the similarity. 'Twas an experiment.

Share this comment


Link to comment
ignoring phase, since it's virtually impossible to control the phase of the TIA audio waveform

 

Is the TIA audio not deterministic, with the main input getting hit exactly twice per scan line? I should think that it should be possible to control the phase precisely on a real 2600 (at a worst-case cost of about 17 scan lines per frame)

Share this comment


Link to comment
ignoring phase, since it's virtually impossible to control the phase of the TIA audio waveform

 

Is the TIA audio not deterministic, with the main input getting hit exactly twice per scan line? I should think that it should be possible to control the phase precisely on a real 2600 (at a worst-case cost of about 17 scan lines per frame)

Oh, yes, it's deterministic, just not easy to control / predict. A collorary would be the digits of pi - the sequence is always the same, but it's impossible to predict the next digit even if you know the previous digits. The problem with the TIA is changing AUDC/AUDF doesn't reset the state of the LSFRs or counter - they continue from the state they were in.

 

The TIA audio generator has three main sections. First is the frequency divisor - a 5 bit up counter which is reset to 0 when AUDF is reached. The problem is if the current value of the counter is greater than AUDF, then it keeps counting up before wrapping around from 31 back to zero. So, to know the phase of the output, we have to track the exact value of the counter.

 

The next two sections are the LFSRs - a 4 bit and a 5 bit which can interact with each other. Again, changing AUDC doesn't impact the state of the LSFRs, but unless the state of the frequency divisor is known exactly, there's little point in trying to track the LSFR state.

 

So for simplicity sake, it doesn't make sense to try to determine the exact phase of the output waveform. For although it might be possible to predict the phase of the output, controlling would be much, much more difficult.

 

And from an desired output it doesn't make much sense either. If the TIA is 90 degrees out of phase from the sample waveform, then it won't match; even though it might be otherwise correct. Finding frequency matches is tough enough without worrying about phase.

Share this comment


Link to comment
Guest
Add a comment...

×   Pasted as rich text.   Paste as plain text instead

  Only 75 emoji are allowed.

×   Your link has been automatically embedded.   Display as a link instead

×   Your previous content has been restored.   Clear editor

×   You cannot paste images directly. Upload or insert images from URL.

Loading...
  • Recently Browsing   0 members

    No registered users viewing this page.

×
×
  • Create New...