A7800 5.1 release

[RevEng]

a7800 v5.1 has now been released at github. 

 

Whats new in this release:

• pokey emulation updated for better two-tone mode accuracy
• fix for cart format supergame+ram+pokey@800 crash

Normally I would have accumulated more changes before a release, but given we have several pokey-focused projects in development (hokey, games using pokey, rmt support for 7800basic) it seemed prudent to get the improvements out quicker than usual.

 

A special thanks goes to @Synthpopalooza for performing in-depth stress testing of the various pokey modes!

 

As usual, @Trebor and I hope you enjoy it! Feel free to use this thread to report any issues, or ask any questions.

 

 

[+Muddyfunster]

Nice work folks. I tested on E.X.O. and it sounds really good. 

[Synthpopalooza]

Needless to say, you all have outdone yourselves on this one.  I am *ALMOST* done with having to cross-develop my music on Altirra. 

[+saxmeister]

@RevEng - many thanks!!!! You guys are making our lives better with each update.

[+bhall408]

Curious - have you ever looked into AtariVox support? While I see some SpeakJet code on GitHub for Arduino, and there is also SpeakJet support in VIDE (Vectrex IDE), I didn't find any in a quick MAME search.

[RevEng]

I haven't looked into it. The speakjet chip is a closed-source pic microcontroller, so emulation would actually require someone to create an open equivalent.

[+bhall408]

A shame it doesn't support plain text strings, in that case it could be "emulated" by using the hosts text to speech facility. Wouldn't sound exactly the same, but could be an interesting experiment.

 

However, perhaps there is a seed of an idea there - a layer that intercepts the allophones and presents them (at that level) to the host speech API.

 

Quote

The AtariVox uses a SpeakJet chip to synthesize speech. Speech is constructed from strings of allophones – the individual sounds that make up speech. Refer to the SpeakJet User Manual PDF from www.speakjet.com for details of the allophone/control codes.

 

[Trebor]

3 hours ago, bhall408 said:

A shame it doesn't support plain text strings, in that case it could be "emulated" by using the hosts text to speech facility. Wouldn't sound exactly the same, but could be an interesting experiment.

 

However, perhaps there is a seed of an idea there - a layer that intercepts the allophones and presents them (at that level) to the host speech API.

The A7800 source code for downloading and GitHub page are open to the public.  External contribution welcomed, especially if they improve or enhance support. 

 

Considering the amount of resources poured into the accuracy of pokey emulation, having something in place that "wouldn't sound exactly the same", would never completely suffice, for more than one reason.  Granted, it would or could be a start though.

[+Karl G]

4 hours ago, Trebor said:

The A7800 source code for downloading and GitHub page are open to the public.  External contribution welcomed, especially if they improve or enhance support. 

 

Considering the amount of resources poured into the accuracy of pokey emulation, having something in place that "wouldn't sound exactly the same", would never completely suffice, for more than one reason.  Granted, it would or could be a start though.

On the 2600 side, the Gopher emulator has added preliminary support for AtariVox voice emulation using a backend of the open-source Festival speech synthesis system. I'm sure accuracy will never be 1 to 1 accurate, but I imagine that it would at least reduce the amount of debugging that would need to happen on hardware by at least testing the speech in a game in broad strokes.

 

Note that I am neither volunteering to add support to a7800, nor suggesting that someone else should do so, but I thought it would at least be interesting to give an example of what is being discussed here.

[+bhall408]

22 hours ago, Karl G said:

On the 2600 side, the Gopher emulator has added preliminary support for AtariVox voice emulation using a backend of the open-source Festival speech synthesis system. I'm sure accuracy will never be 1 to 1 accurate, but I imagine that it would at least reduce the amount of debugging that would need to happen on hardware by at least testing the speech in a game in broad strokes.

That is a conceptual approach we were discussing (Argon devs) earlier this week -- translate the SpeakJet requests into one the (Android) OS provided TTS system can use.

 

[+bhall408]

22 hours ago, Karl G said:

Note that I am neither volunteering to add support to a7800, nor suggesting that someone else should do so, but I thought it would at least be interesting to give an example of what is being discussed here.

For the record, we have decided to have a go at switching from the MAME 7800 core to A7800 in a future release of Argon... After that, we'd be in a position to be helping out with enhancements...