SOUND : GSV-R v1.1

[JamoZ]

And they still sounded horrible

[HornetMaX]

Hmm ... how can you get a sample of an engine running sufficient time at constant rpm off-gas ?

For SuperbikeX the Engine sound Samples were taken from a dyno session with each bike several Times.

Hmmm ... can the dyno keep the wheel (and hence the engine)  spinning at constant speed ?

Anyway, some partially useful info on sound creation here:

http://www.nogripracing.com/forum/showthread.php?t=21694&highlight=sound+tutorial
http://isiforums.net/f/attachment.php?attachmentid=139&d=1288395585

As soon as I'm done with everything else, I may want to have a look at this sound stuff ... the quality of the current sounds (except original ones) is below the rest.

MaX.

[JamoZ]

I disagree. The stock Murusama sound is horrible and sounds more like a coffee machine then a motorcycle engine. The 2 strokes are decent, but not great aswell. The sound engine is just not sufficient. If you take a look at AC or rFactor`s sound engine they allow alot more samples to be used for 1 car. Could be up to 30 samples while we only can work with 11...

[Stout Johnson]

I disagree. The stock Murusama sound is horrible and sounds more like a coffee machine then a motorcycle engine.

I  think MaX refers to the Murasama sound at least sounding consistent and the samples run smoothly into each other...

[HornetMaX]

I disagree. The stock Murusama sound is horrible and sounds more like a coffee machine then a motorcycle engine. The 2 strokes are decent, but not great aswell. The sound engine is just not sufficient. If you take a look at AC or rFactor`s sound engine they allow alot more samples to be used for 1 car. Could be up to 30 samples while we only can work with 11...

I really doubt the difference is due to 30 samples instead of 11 samples. As far as I can see, most of the games have a very very similar approach to engine sounds.

The default sounds at least do not have the following problems (almost always present in mod sounds):

• you can hear two distinct sounds when there is a transition from one sample to th eother
• sometimes, even within one sample you can hear multiple "layers"

After that it all become personal opinion: some prefer an horrible sound, just to have a different sound from the default bikes. I personally cannot stand the "two-overlapping-sound" issue.
It's terribly present in your video above (and in almost all mod sounds I've heard to date): it sounds like another bike is somewhere behind you.

Of course that's not your fault: even forgetting about the lack of documentation, creating a good sound is probably very tough. No wonder Piboso delegated this to a specialist at the time ...

MaX.

[iVolution]

Exactly, is imposible, you have to edit-trial-aww sounds awfull, edit-trial- aww worse, edit-trial- mmm maybe , edit-trial- F**K this S**T, until you have something at least similar to the real thing.

Good to see you putting you thought process on paper, it always is the first step to recovery 

[Warlock]

Hahaha, i'm sorry its really difficult for me to explain this kind of things in english. Also im an absolute noob in sound editing.
Anyway i mostly do all this stuff by ear and trial-error. The tools we have to work with are not enough (engine.scl)
Also sound engine works in a way you always have 2 layers of sound.
Max, try this, rename all on_throttle sound files, so you cant hear them. Go and try it ingame. Result:

if you do all the track at mid throttle you will still hear off_throttle sounds, ....and you are ON_throttle.
So the sound engine mixes on and off throttle, depending of the different pitches of both sounds, almost always sound very weird 2 layers sound, it tends to be robotic.

So if you create a 'great' on_throttle sound you think "bingo!", then you go ingame and sounds awfull because is mixed with off_t sounds, frustrating  .

Also we render in mono and kills some details of the 'great' sample we've made, not to mention searching for good samples 
You wouldn't believe where i got the sample im working on now for 1000cc bikes  ............... a 4T Radio Control plane engine LOL  , with some heavy tuning in a sound editing software and some praying to the Virgin Mary it turned out quite good  . Still plenty of work to be done, so not showing it yet.

Source for the sample:
https://www.youtube.com/watch?v=rKY3JA8Y4rA

etc, etc , etc , its really a huge effort try to do something that sounds at least similar to a real bike

[SwarleyRuiz]

great job!

[C21]

Hmmm ... can the dyno keep the wheel (and hence the engine)  spinning at constant speed ?

Hi Max.
Yes, the dyno can. (not every dyno of course!)
That´s how a map is done on a real bike ;-)
Thrust me 
http://www.youtube.com/watch?v=JtwrJCGx9a8

[HornetMaX]

Also sound engine works in a way you always have 2 layers of sound.

... ... ...

So the sound engine mixes on and off throttle, depending of the different pitches of both sounds, almost always sound very weird 2 layers sound, it tends to be robotic.

Warlock I know the sound engine always mixes two sounds (on/off throttle), Piboso explained that once (and it's even explained in one of the links I posted above: may not be exactly as GPB does it but the concept can't be far off).

Ask yourself this question: on the 990 sound, do you hear 2 layers ? The answer is a clear NO. Never, no matter the situation.

What does this teach ? That the samples of the 990 have some particular property that makes them work well with the blending (on/off) and transitioning (from one sample to the next/previous) mechanisms.

I was thinking about this this morning and I'm about to do some testing to validate my reasoning: I think two potential problems nobody has seen may be that overlapping samples (i.e. consecutive samples in the same layer + on/off samples at the same RPM)  ay need to be very close in terms of pitch where they overlap and may also need to be "in phase".

I'm really not sure of the above, that's why I'll do some basic testing with pure tones.

Hmmm ... can the dyno keep the wheel (and hence the engine)  spinning at constant speed ?

Hi Max.
Yes, the dyno can. (not every dyno of course!)
That´s how a map is done on a real bike ;-)
Thrust me 

I do trust you but I'm not sure I understand what you mean with "map".

MaX.

[C21]

I do trust you but I'm not sure I understand what you mean with "map".

remap of the ecu.

[HornetMaX]

I do trust you but I'm not sure I understand what you mean with "map".

remap of the ecu.

Hmm that does not enlighten me a lot I'll PM you or I'll be seriously off  topic here (I know, I'm a bad example).

MaX.

[HornetMaX]

So, I'm finally done with the GPB guide and hence I have time for other stuff. Before diving into physics modding (a dive I may never surface from), I'm having a look at the sound stuff.

@Jamoz/Warlock: a few comments on the GSVR sound:

• There are some very long samples (13s): usually 2-4s are enough.
• There are some samples that seem to be identical (e.g. sb2_idle, sb2_offidle and sb2_onidle): this is redundant, you can just use one single file.
• You are not obliged to name the samples as the original ones: it's better to have different names. This way you can have engine1.scl referencing some samples and engine2.scl referencing some other samples. You will be able to switch between the two simply editing sfx.cfg (and pointing to engine1.scl or engine2.scl).
• I'm under the impression you think in GPB we are obliged to use 11 samples: that's not the case, you can edit the .scl file and remove samples. Likely you can also add samples to use more than 11, but this is TBC, only Piboso knows if there's an upper limit (I'm too lazy to try to figure out).

One question concerning your GSVR sound: how did you proceed to create the engine.scl ?
I've found most of the values are identical to the 990's engine.scl: given that you used different samples, this is unlikely to be correct.

From the first tests I've done, it seems that it is very important to know the RefPitch of your sample: the RefPitch multiplied by MaxRPM (e.g. 17000 for the 990) gives you the number of RPMs at which your sample will be played without any pitch change (i.e. as it is). In other words, the RefPitch of a sample is the number of RPMs at which the sample has been taken (divided by the MaXRPM). If you get this wrong, things tend to go very wrong.

How did you get the RefPitch for your samples ?

MaX.

[JamoZ]

Max, tbh i have no idea. I just do the sound stuff for fun and to have at least a different sounding bike. For the base i think i copy`d warlocks R6 engine file and with some help of stout i just upped the volume levels and pitch levels. Don`t remember doing anything to the refpitch value`s.

This info will sure be very handy for the guys who are doing alot of effort to produce some very good sounds.

I just wish one day we could have sounds close to this in gpb`s (skip to 55 seconds to hear that sweet lotus 79)

http://www.youtube.com/v/C9zQ-dSlQ_g#t=49

[HornetMaX]

Max, tbh i have no idea. I just do the sound stuff for fun and to have at least a different sounding bike. For the base i think i copy`d warlocks R6 engine file and with some help of stout i just upped the volume levels and pitch levels. Don`t remember doing anything to the refpitch value`s.

Zero chances to get something half-decent proceeding this way. Well, I guess it depends what we call half-decent

From what I see, GPB sound engine is rather capable so it should be possible to get something nice. But don't forget we may have no access to the tools professionals use for that kind of result (recording hardware, access to real models, sound gurus etc).

MaX.