Stella audio settings? What does each thing do?

[+Random Terrain]

I Googled and looked at the included documentation, but I don't see an explanation for any of this:

 

Volume - This seems easy to understand.

 

Fragment size - What does this do?

 

Output freq - What does this do?

 

TIA freq - What does this do?

 

Clip volume - What does this do?

 

Enable sound - This seems easy to understand.

[+stephena]

I Googled and looked at the included documentation, but I don't see an explanation for any of this:

 

Volume - This seems easy to understand.

 

Fragment size - What does this do?

 

Output freq - What does this do?

 

TIA freq - What does this do?

 

Clip volume - What does this do?

 

Enable sound - This seems easy to understand.

The middle ones are mostly technical, and aren't really meant to be messed with (unless you're having problems). A quick explanation follows:

 

1) Fragment size: Determines the size of a buffer into which the sound is created. Roughly speaking, the smaller the buffer size, the more 'in-sync' the sound will be with the video, but at the expense of more CPU usage (and some sound cards can't handle too low a buffer size at all). Making this buffer larger 'fixes' such issues with sound cards, but introduces lag (ie, there may be a noticable delay between what you see onscreen and when you hear a sound associated with it).

 

2) Output frequency: Ideally, this will be the same as the original frequency of the TIA sound hardware (31400Hz). It's basically the frequency at which the sound card is opened for playing sound.

 

3) TIA frequency: This one is the frequency used by the internal TIA sound emulation. It should also ideally be 31400Hz, and more specifically, the same as the output frequency. Things don't sound too well if you mess with the frequency options, but on some sound cards it's the only way to get any sound at all.

 

4) Clip volume: Essentially cuts the range of volume output in half. If this is enabled (the default), it eliminates annoying cracking/popping sounds when pausing emulation, switching from emulation to debugger/launcher modes, etc. Disabling it obviously doesn't fix the popping issue, but it would allow the volume to much louder (ie, a volume setting of 100 in clipped mode is roughly equal to 50 in non-clipped mode).

 

Anyway, that's pretty much it for the sound stuff. My advice is if things are working wrt sound, you probably shouldn't mess with these items.

[+Random Terrain]

Thanks for the explanation.

[yuppicide]

It's an old post, but I thought it deserved a bump since it answered my question.

 

I had Stella 3.3, but only recently started fooling around again since building this new PC. I had sound issues with some popping noise. Thought maybe it was my game code.

 

Downgraded to Stella 3.2 with same issue.

 

I found out it was a Fragment Size issue in my audio options. I fooled around and 4KB worked for me.

[+Random Terrain]

I had a problem similar to the one that yuppicide had. Ever since I got my Windows 7 computer, sound in Stella has been randomly dropping out and generally sounding like crap. Under Audio Settings, I changed Sample Size to 1 KB, left Frequency at 31400 Hz, closed Stella, then opened it again. That seemed to fix the sound problem and there doesn't seem to be any lag at a sample size of 1 KB. There was lag when I tried 2 KB and 4 KB, though.

 

Speaking of sound, maybe one day the sound that emulators produce will sound exactly like a real Atari 2600 (for example,

).

[Trebor]

I had a problem similar to the one that yuppicide had. Ever since I got my Windows 7 computer, sound in Stella has been randomly dropping out and generally sounding like crap. Under Audio Settings, I changed Sample Size to 1 KB, left Frequency at 31400 Hz, closed Stella, then opened it again. That seemed to fix the sound problem and there doesn't seem to be any lag at a sample size of 1 KB. There was lag when I tried 2 KB and 4 KB, though.

 

Speaking of sound, maybe one day the sound that emulators produce will sound exactly like a real Atari 2600 (for example,

).

 

Sound "lag", "dropping out", and "generally sounding like crap" at different sample size for different systems is largely impacted by the system itself including speed, memory, and driver performance. 4KB worked for yuppicide, 1KB works for Random Terrain, and for me 512 bytes is best. I'm thankful stephena provides the choice of sample size instead of telling people they need a better computer or/and to go complain to the Creative/Microsoft to fix their drivers.

 

The audio out from a recent version of Stella, while also trying the variety of Audio options provided, needs to feed through the same speakers the real system is being fed out of in order to make a valid recent comparison. Additionally, the video linked is old, showcasing a very old version of Stella from 5+ years ago. The sound in Stella has improved since Spring 2008.

 

This reminds me of people stating that under MAME how Donkey Kong doesn't sound like the arcade and forgetting the physical conditions that surround what they hear in the Arcade (Including speaker quality/enclosed-echo of cabinet/etc.). Set a MAME machine outputting to an original Donkey Kong cabinet with the original working speakers and then you will hear how well it emulates the real machine.

 

A speaker's frequency response, signal to noise ratio, and overall physical hardware build quality are just some of the things that will affect the end results. Take any music source and feed it through a pair of $10-40 speakers and then feed that same source through some Kilipsch, Paradigm, or similar quality speakers, and you will hear a difference in sound quality output as well.

 

If able/possible try changing the low end frequency levels and overall bass output of, or to, the speakers you are utilizing with Stella. It may assist in reaching a better match to the actual hardware from what was recorded capturing Stella in that old clip if you are experiencing similar results in a recent version.

[+Random Terrain]

Sound "lag", "dropping out", and "generally sounding like crap" at different sample size for different systems is largely impacted by the system itself including speed, memory, and driver performance. 4KB worked for yuppicide, 1KB works for Random Terrain, and for me 512 bytes is best. I'm thankful stephena provides the choice of sample size instead of telling people they need a better computer or/and to go complain to the Creative/Microsoft to fix their drivers.

 

The audio out from a recent version of Stella, while also trying the variety of Audio options provided, needs to feed through the same speakers the real system is being fed out of in order to make a valid recent comparison. Additionally, the video linked is old, showcasing a very old version of Stella from 5+ years ago. The sound in Stella has improved since Spring 2008.

 

This reminds me of people stating that under MAME how Donkey Kong doesn't sound like the arcade and forgetting the physical conditions that surround what they hear in the Arcade (Including speaker quality/enclosed-echo of cabinet/etc.). Set a MAME machine outputting to an original Donkey Kong cabinet with the original working speakers and then you will hear how well it emulates the real machine.

 

A speaker's frequency response, signal to noise ratio, and overall physical hardware build quality are just some of the things that will affect the end results. Take any music source and feed it through a pair of $10-40 speakers and then feed that same source through some Kilipsch, Paradigm, or similar quality speakers, and you will hear a difference in sound quality output as well.

 

If able/possible try changing the low end frequency levels and overall bass output of, or to, the speakers you are utilizing with Stella. It may assist in reaching a better match to the actual hardware from what was recorded capturing Stella in that old clip if you are experiencing similar results in a recent version.

 

The age of the video doesn't matter. The ship landing still basically sounds the same as it did before. I'm also using speakers from a stereo, not crappy little PC speakers. My Atari 2600 is hooked up to the same speakers, so that's not the problem. The ship landing sound effect also doesn't sound right using the newest emulator out there (JAVATARI). It can't be blamed on a crappy sound card or crappy speakers this time. Something isn't quite right with the sound in every Atari 2600 emulator I've tried so far. There must be some little thing that is being overlooked. Once someone figures out what needs to be adjusted, maybe emulators will finally sound like a real Atari 2600. They are close in most cases, but certain sound effects really make the differences jump out at you.

[Trebor]

The age of the video doesn't matter. The ship landing still basically sounds the same as it did before. I'm also using speakers from a stereo, not crappy little PC speakers. My Atari 2600 is hooked up to the same speakers, so that's not the problem. The ship landing sound effect also doesn't sound right using the newest emulator out there (JAVATARI). It can't be blamed on a crappy sound card or crappy speakers this time. Something isn't quite right with the sound in every Atari 2600 emulator I've tried so far. There must be some little thing that is being overlooked. Once someone figures out what needs to be adjusted, maybe emulators will finally sound like a real Atari 2600. They are close in most cases, but certain sound effects really make the differences jump out at you.

 

It doesn't eliminate driver issue(s), equalizer (lack of) settings/response on the sound card's output, sound card (in)capabilities and other misc. system settings. Certainly the sound card is not being blamed, but cannot be overlooked or exclude from the equation; regardless that both the system and your PC is outputting to the same speakers, PC soundcards vary in quality and the type of sound they put out at different settings and frequencies. Especially when going outside of the realm of a rate of 44100 or 48000, many can start having issues with some or many effects.

 

RT, I'm not saying Stella sound is perfect, but using a 5 year old video reference without all variables identified does not serve as evidence that the emulators (especially current releases) are not matching a specific sound or series of sounds. Furthermore, there is more involved and at play than just connecting the system and a computer to the same set of speakers. It covers some variables not all.

 

On the other hand, there has been talks by some of even going down to gate level emulation of the 2600; however, even if that ever comes to fruition, I'm sure there will be someone(s) saying something isn't "right" (Again, I reference the aforementioned Donkey Kong audio claims). Audio much like video, can vary greatly depending on all hardware components and factors involved. What one person claims is audio and video perfection from what they experience(d) or is/was used to can be "wrong" to another person.

 

Nonetheless, I did locate this little bit which may explain any audio deficiency with Stella and Atari 2600 emulators in general:

http://atariage.com/...00#entry2708930

 

More on frequency factor specifics here:

http://atariage.com/...00#entry2709043

[SeaGtGruff]

2) Output frequency: Ideally, this will be the same as the original frequency of the TIA sound hardware (31400Hz). It's basically the frequency at which the sound card is opened for playing sound.

 

3) TIA frequency: This one is the frequency used by the internal TIA sound emulation. It should also ideally be 31400Hz, and more specifically, the same as the output frequency. Things don't sound too well if you mess with the frequency options, but on some sound cards it's the only way to get any sound at all.

Should these be set to 31113Hz for PAL/SECAM emulation, since that's the half-line frequency for the TIA on PAL/SECAM consoles? Or does Stella make some sort of internal adjustment for the different oscillator rate when emulating PAL/SECAM games? All you'd need to do internally is multiply the NTSC half-line rate (approximately 31400Hz) times 3546894 / 3579575 (PAL oscillator rate divided by NTSC oscillator rate) and round the result, giving 31113Hz.

[+Random Terrain]

It doesn't eliminate driver issue(s), equalizer (lack of) settings/response on the sound card's output, sound card (in)capabilities and other misc. system settings. Certainly the sound card is not being blamed, but cannot be overlooked or exclude from the equation; regardless that both the system and your PC is outputting to the same speakers, PC soundcards vary in quality and the type of sound they put out at different settings and frequencies. Especially when going outside of the realm of a rate of 44100 or 48000, many can start having issues with some or many effects.

 

RT, I'm not saying Stella sound is perfect, but using a 5 year old video reference without all variables identified does not serve as evidence that the emulators (especially current releases) are not matching a specific sound or series of sounds. Furthermore, there is more involved and at play than just connecting the system and a computer to the same set of speakers. It covers some variables not all.

 

Will you drop the 5 year old video stuff? I linked to it because that's how the ship landing sound effect still sounds. I'm not going to record a special new video for a sound effect that hasn't changed.

 

I'm using a Sound Blaster X-Fi Titanium. The audio control panel has many tabs. Do you know what I should be adjusting?

 

 

 

 

 

What one person claims is audio and video perfection from what they experience(d) or is/was used to can be "wrong" to another person.

 

Yeah, probably because most people's ears are similar to their tongues. There are super-tasters, medium-tasters, and non-tasters. It seems like most people are either 'non-hearers' or 'medium-hearers.' They are the people at shows like American Idol who boo the judges for pointing out that a singer was off-key and sounded like someone shoved a live chicken down their throat. To most non-hearers in the audience, the singer sounded wonderful.

[Schizophretard]

I'm confused. Lets say that RT's audio settings, sound card, drivers, and whatever on his computer are the problem and the audio of Stella is accurate. Then why would RT's computer ,while making the video, make the sounds from the original hardware sound accurate while making Stella sound different? On top of that, why would his video going through my computer's audio still make his original hardware sound accurate? I'm two computers and the internet away from RT's Atari but I hear an Atari. How is RT able to make sounds exit my headphones that match the sounds that would come out of my Atari but with the same computer he gets different results with Stella that are similar to my own results with Stella?

[+stephena]

I can sum up this conversation quickly. There is indeed still a problem with the sound emulation in Stella. And it's there in all other emulators too, since they either (a) use Ron Fries TIA sound code, or (b) derive their code from the incomplete documentation out there.

 

I've said several times in the past that I'm not a sound person, and I'm not well versed in the intricacies of sound programming. At some point, some people (probably including me) have to get together and research this more fully. But my time is limited ATM, and I'm concentrating more on the graphical parts of the TIA emulation.

 

EDIT: I should add that while I'm reasonably sure the problem is in the sound code emulation, I can't say if that's the only thing that's causing the problem(s). I haven't researched this enough to really say anything definitive, but I can say that all the 2600 emulators seem to have the same issues, which leads me to think that it is indeed an emulation issue, probably cause by incomplete documentation.

[SeaGtGruff]

the incomplete documentation out there.

I can help with the documentation, at least with the TIA aspects of it (but not so much with any high-pass, low-pass, or band-pass filtering that gets applied after the AU0 and AU1 signals leave the TIA chip). I have a set of spreadsheets I made that simulate the TIA's sound generation on essentially a gate-by-gate basis. I've been planning to write up a document that would explain it, and include the spreadsheets.

 

all the 2600 emulators seem to have the same issues, which leads me to think that it is indeed an emulation issue, probably cause by incomplete documentation.

I'm reasonably sure the existing sound emulation is probably "too simplistic" in two or three different ways. First, I presume the code is designed in such a way that changes to AUDF0/1 and AUDC0/1 have virtually instant effects, whereas changes to those registers on actual consoles need time to "propagate" (if that's the correct term here). Changes to AUDF0/1 can take the most time, because the frequency counter typically needs to finish counting the old frequency before the new frequency takes effect. But changes to the AUDC0/1 registers also need time to propagate, at least for some of the values, because the shift registers need to shift out any current values before the new values can be output-- and since the shift registers are clocked by the output from the frequency counters, that might take a while. Second, I don't think the emulation tries to replicate the effects of any filtering that gets applied to the audio signals after they leave the TIA. I don't know how much difference that makes. And third (and probably least important), the audio emulation probably isn't timed to be in step with the scan lines the way the TIA is-- that is, the two audio clock signals are triggered four times (twice each) at specific times on a scan line. So the effect of writing to AUDF0/1 or AUDC0/1 may not even begin the propagation process until as much as half a scan line later when the next audio clocks are triggered. I presume this third consideration has a negligible impact, whereas the impact of the first and second considerations (in that order) are potentially more significant.

[+Random Terrain]

I can sum up this conversation quickly. There is indeed still a problem with the sound emulation in Stella. And it's there in all other emulators too, since they either (a) use Ron Fries TIA sound code, or (b) derive their code from the incomplete documentation out there.

 

I've said several times in the past that I'm not a sound person, and I'm not well versed in the intricacies of sound programming. At some point, some people (probably including me) have to get together and research this more fully. But my time is limited ATM, and I'm concentrating more on the graphical parts of the TIA emulation.

 

EDIT: I should add that while I'm reasonably sure the problem is in the sound code emulation, I can't say if that's the only thing that's causing the problem(s). I haven't researched this enough to really say anything definitive, but I can say that all the 2600 emulators seem to have the same issues, which leads me to think that it is indeed an emulation issue, probably cause by incomplete documentation.

 

With the help of people like SeaGtGruff, Stella might be the first Atari 2600 emulator to sound exactly like a real Atari VCS. Then the other Atari 2600 emulators will eventually improve their sound too.

[Trebor]

I'm confused. Lets say that RT's audio settings, sound card, drivers, and whatever on his computer are the problem and the audio of Stella is accurate. Then why would RT's computer ,while making the video, make the sounds from the original hardware sound accurate while making Stella sound different? On top of that, why would his video going through my computer's audio still make his original hardware sound accurate? I'm two computers and the internet away from RT's Atari but I hear an Atari. How is RT able to make sounds exit my headphones that match the sounds that would come out of my Atari but with the same computer he gets different results with Stella that are similar to my own results with Stella?

 

Sorry, Schizo...Didn't mean for this to get confusing. My initial point was aimed at stating that pointing to a five year old video clip from a five year old version of Stella and stating it doesn't sound like the original hardware would not be the best form of evidence for recent issues or problems with Stella's sound emulation. I also tried to highlight how there are other factors which impact the end result of what people hear from the same sound source. Similar to the way video differences works as well. For example, the same DVD or Blu-Ray can look and sound different on a variety of setups; especially if similar to this Stella issue just homing in on one or two specific background noises or frequencies played back.

 

I absolutely agree, Stella sound is not perfect. And yes there is indeed some fixes needed. That is why I posted the quotes here regarding the recognition of the need for improvements by stephena, but it just was not the focus right now (TIA graphic fixes is the focus). SeaGtGruff may be able to assist with the sound improvements though and just like the group effort that has been exerted in tackling the graphical bugs, hopefully the sound bugs can be conquered similarly. Sorry RT too for any miscommunication(s) on my part.