Space Taxi - A8B version

[Jacques]

23 hours ago, Andreatari said:

currently the samples are good enough to play the game but their quality is lower compared to C64 version due to the double step I used: I've recorded an already digitized voice and I did it again.

Did you use digital output (writing to wav file) of the C64 emulator for that purpose?

[+MrFish]

6 minutes ago, Jacques said:

Did you use digital output (writing to wav file) of the C64 emulator for that purpose?

WinVice can record WAV sound files. It should produce pretty good results.

 

[miker]

Or maybe take some available voice sample, such as this:

https://movie-sounds.org/old-movie-samples/quotes-with-sound-clips-from-casablanca-1942/hey-taxi

 

(could be manipulated a bit, though)

 

Edit: or use such a site to to all of the voices maybe: https://www.narakeet.com/languages/

 

Edit 2: used American accent and voice "Mike" for example.

Edited July 25, 2023 by miker
update

[Andreatari]

Hi @Jacques and @MrFish,

I used a .sid file containing all sounds and a sid player but probably the digital output of WinVice could give better results... the phrases are reproduced at different pitches and I need all sound at the same pitch (than I can change the speed in the program to get different tones just like Kutcher did in the original game).
I'll make some testings with WinVice outputs... thanks for your suggestion 👍

Hi @miker
using the original sounds we lose quality, it's true... also because the phrases from the original game have been already sampled and, despite the quality of the wav file, I have to resample it to 3889Hz, convert it to encoding-unsigned 8 bit PCM and clean the header with an hex editor before using the procedures suggested by Janusz 😬

A possible solution could be to use some text-to-speech procedure to get an inital better quality, however I think it's very important to keep the same 'feelings' of the original Space Taxi rather than equipping it with a brand new voice... Maybe I'm wrong but the testings I made with new recordings were definitely unsatisfactory and a bit strange 😝


By the way... I'm working on custom, simple compression routines in my Atari version.
The target is a real 130k disk and the first results are going in the right direction 😎
 

Edited July 25, 2023 by Andreatari

[+MrFish]

52 minutes ago, Andreatari said:

I'll make some testings with WinVice outputs...

WinVice shows the following options for saving sound files.

 

 

[Wrathchild]

you can hear the samples on the Lemon site page for the game but the original doesn't seem so clear anyway?

[Andreatari]

Hi All,

 

Just an update about the Space Taxi work in progress.

Let's start with the bad news... I don't see many solution about the quality of the samples 😕
I made some testings with VICE 64 exported WAV files but, despite a better starting quality of the file, the final result after the resampling and preparation for the game, is more or less the same.
Just to be sure, I tried with a brand new recording and the result is completely different.
Here attached you can hear 3 WAVs:

1 - C64 exported WAV, just as you can hear from C64 - NOT resampled - 124K
 

 

2 - A8 WAV resampled at 3889 Hz, ready to be imported in the game - 5K
 

 

3 - My voice recorded with a smartphone and resampled at 3889 Hz, ready to be imported in the game - 5K
 

 

I know that my voice is not the best and it's not good for the game and it's obviously different from the original, but it helps to understand how the quality could be increased.
So now we have a couple of options... to keep the current voices with lower quality but closer to the original (my preferred choice) or record some good new voices and get a better quality.
What's your opinion?

And now some good news 😎
I'm applying minor changes and bug corrections (solved the bad "not-existing destination Pad" bug) but the main news is that
- I've made several changes to save RAM
- Ive compressed almost all files with a custom routine, simple but fitting the program structure perfectly
- I've changed the loader and I've added a small routine to expand data, so...
currently the whole game fits in 130k DD disk readable by any standard 1050 drive 👍

 

Before the middle of August I'll release this 2.5 version and I'll try to understand if it wil be possible to prepare also a cartridge version of the game (thanks @Wrathchild for your help 🤗... I've other questions for you 😬)
 

That's all at the moment

Ciao

 

Andrea
 

[+Stephen]

11 hours ago, Andreatari said:

Before the middle of August I'll release this 2.5 version

If you release on the 12th, feel free to make it a release for me

[Irgendwer]

On 8/2/2023 at 5:14 PM, Andreatari said:

What's your opinion?

Maybe you could give S.A.M. a chance?

 

Variation of tune, expression and speed should be easily possible for different passengers then.

 

AFAIR @Heaven/TQA is experienced in using it and could give some useful hints...?

 

Edited August 3, 2023 by Irgendwer

[Jacques]

For me SAM's speech is a goid idea, it's style is very 8-bit, and foremost, it could produce better quality. Original is original, but when it's hard to understand or it doesn't do justice Atari capabilities, it could be replaced, methinks. 

[Andreatari]

Hi all and thanks for your feedbacks 😊

Honestly I'm open to any solution and I understand the reasons why could be convenient to follow other ways... As I told, I would like to release a DD disk 'final version' about mid August and it would be very interesting to have a couple of alternatives to the current voice synthesis.

Currently I'm working on new samples using Narakeet Mike voice with american accent as suggested by @miker but I could also try to use SAM as proposed by @Irgendwer and @Jacques.
The problem with SAM is that I absolutely have no experience with this system so I'd have some question:

- Does it exist a stand alone SAM routine able to speak the phrases that could be easily fitted in one (or two) 4K memory bank with all the necessary data?
- Does SAM need any HW device or special feature or it works on a simple, standard XL/XE machine?
- During SAM speech, could we see the screen with DLIs and a short VBI (in 1 level) working or we could have a sort of 'black screen'?
- In the example proposed by Irgendwer I don't understand the second word... are we sure that the SAM quality would be really so better? 🤔

Remember also that in any case the voice would be available only in expanded machines because I' really have no other free memory anymore 😕
 

Ciao

 

Andrea

[Rybags]

I think SAM lives in low memory just above where Dos normally loads.

It normally blanks the screen but has the option not to, though the quality drops significantly.

A big part of that problem is because it uses delay loops.  Ideally it would be good to replace that with Pokey Timers as they would reduce the jitter even with active screen and DLIs going on.

[_The Doctor__]

just recording better samples should suffice, maybe you can grab them straight from video, movie, tv whatever.... TAXI!!!!!

PAD number...

 

[Jacques]

11 hours ago, Andreatari said:

Currently I'm working on new samples using Narakeet Mike voice with american accent as suggested by @miker but I could also try to use SAM as proposed by @Irgendwer and @Jacques.

 

Cool, thank you! The American accent should do well, if it's much clearer than original in the end, perfect

[Irgendwer]

18 hours ago, Andreatari said:

The problem with SAM is that I absolutely have no experience with this system so I'd have some question:

- Does it exist a stand alone SAM routine able to speak the phrases that could be easily fitted in one (or two) 4K memory bank with all the necessary data?

I'm not sure if 4k is enough. Inspecting the .obj-file inclusion here:

...I would guess more memory is needed.

18 hours ago, Andreatari said:

- Does SAM need any HW device or special feature or it works on a simple, standard XL/XE machine?

No external hardware required. The solution is fully software based.

 

18 hours ago, Andreatari said:

- During SAM speech, could we see the screen with DLIs and a short VBI (in 1 level) working or we could have a sort of 'black screen'?

It's possible to leave the screen on while outputting speech. However I'm not fully aware about the restrictions then.

 

18 hours ago, Andreatari said:

- In the example proposed by Irgendwer I don't understand the second word... are we sure that the SAM quality would be really so better? 🤔

Hmm - I find it quite understandable. You are able to vary the output a lot by using different phonemes, pauses etc. and even have control over other parameters.

 

Playing around with this:

https://imrane03.github.io/better-sam/

might be insightful.

[Heaven/TQA]

sorry. late on the party.

SAM is easy to use. grab the manual at Atarimania. there is all information needed for using it in ASM.

[+DjayBee]

On 8/6/2023 at 10:34 PM, Andreatari said:

Does it exist a stand alone SAM routine able to speak the phrases that could be easily fitted in one (or two) 4K memory bank with all the necessary data?

I just had a quick look at its memory consumption using ? FRE(0) in BASIC.
21896 when SAM is loaded

32274 without SAM

 

So it uses a little more than 10K of which the loaded code are 9607 bytes ($2000-$4586).

[_The Doctor__]

I think the samples are the way to go over SAM, unless you sample SAM after you tweak the crap out of the phonemes!

[+DjayBee]

36 minutes ago, _The Doctor__ said:

unless you sample SAM after you tweak the crap out of the phonemes!

I understood the idea as tweak the crap out of the phonemes and have SAM speek "live".

[+CharlieChaplin]

2 hours ago, _The Doctor__ said:

I think the samples are the way to go over SAM, unless you sample SAM after you tweak the crap out of the phonemes!

 

With MPG-8 compression we can tweak SAM to less than 2KBytes, great isn't it ?!?

There is only the small problem that the MPG-8 decompressor takes a few megabytes of RAM... 

 

So, I would use the samples as they are and simply concentrate on creating a Space Taxi version that works from 130k or at least 180k disks/images.

[_The Doctor__]

Fix the speech issue in post!

but that can confuse isn't post the power on self test? inner geek giggles!

[Andreatari]

Here we are... 👍

After several weeks of hard and hot (up to 39° celsius in Italy 🥵 ) work, I can finally release version 2.5 of Space Taxi for Atari 8 Bit.

🚕
 

This could really be considered a kind of "final" version, if we don't want to think about a possible cartridge version.

As usual, You can find Space Taxi 2.5 in the @Philsan's great website (Grazie Filippo!):

https://www.santellocco.com/atari/spacetaxi.htm

 

I have fixed several bugs or uncorrected behaviors, but more importantly, I have introduced some important features and new features. You can read the full changelog in the version history text file included in the downloadable zip but, as a brief summary, these are the main new features:

24/08/2023 - Space Taxi 2.5  - 130K disk version 💪

• - Most of the files have been compressed with a simple custom routine that fits the program perfectly.
• - The boot program has been modified by adding a decompression routine stored in low memory.
• - New voices! All phrases have been resampled using a TTS engine for better quality and splitting words to save space.
• - Classic voices saved for the nostalgic... I prepared an identical copy, but using the original samples. It could be a perfect "Side B" of the floppy
• - Added "Hey" phrase when the cab hits the customer
• - Added phrase "Hey Taxi" when the cab lands on a different pad than the one the customer is on
• - Added new phrase "Game Over" at the end of the game
• - Added the phrase "Pad + No." when the cab takes the customer to the wrong pad. The ride bonus is also reset in this case
• - Added level name to the menu in cheat mode for easier choice
• - Added music for the high score writing screen when the player earns more than $50 and reaches one of the top ten positions  👍
• - Added music for hi-score table 👍
• - Fixed flickering of rainbow bar colors in Level 20 - Fast Break during jingles and during speech synthesis

As always, also for this version I have tried to follow your suggestions and have taken your comments into great consideration... in the end, I achieved almost all possible goals (better quality of samples, the game fits into a 130k disk, sounds and music completed) and the game is stable, but eventual bugs are always possible... for the future I will only work on any major bugs and in a possible (but not easy) cartridge version.

I really hope you enjoy the game and I am sorry that it is not perfect but, after all, nobody is 😎

Thank you all for your support

Andrea

 

[_The Doctor__]

What's to be sorry for, what's not perfect?

You did a great thing, and like almost anyone that's produced something for the Atari, eventually all the things you want to tweak will be in time... even 5 10 20 years later

 

[Jacques]

23 hours ago, Andreatari said:

I really hope you enjoy the game

And I really do, thank you so much! New voices are stunning: clear, nice and somehow they take the joy of playing to the new level 🤩

[Andreatari]

Thank you all for your reactions and feedback... if you want to give a look to any level out of 25 available, you can see this longplay video of the full 24H shift 😎

Space Taxi A8 - Full 24H Shift - Ver. 2.5 PAL

 

Edited August 26, 2023 by Andreatari