Hey Tim, can you spill the beans on anything about new addiions to the sound engine? Would absolutely love a place in the SFX file for turbo boost/spool/blow-off sounds instead of having to mix them with the main engine/gear sounds, cheers.
The main problem which detracts from the "experience" of sound in racesimulations is normalisation. It has been like this since the beginning of racing software. In a real racecar the difference between the engine in the garage at 4000rpm and at 14000rpm is 70dB. A V12 racing car revving to 14000rpm in a closed garage is one of the most powerful experiences ever. In rfactor (1 or 2) the difference is 0dB, the only thing that changes is the pitch of the sound. That is the primary reason that racing games feel "weaker" / less impressive, than real life. That same V12 now feels like a computer game. This is because the simulation software normalises sound dB output range to standard amplifying devices (little weak speakers, headphones, etc). There should be an option to change to a Real / High-Fidelity sound mode. This can be calibrated by setting your maximum desired sound output per session, and scaling back the simulated dB sound output at idle rpm's accordingly. Instead of normalising. (or if it's too much, giving the option of a factor 1.5x, 3x, 6x, etc). This way those that play on an amplified stereo or high-fi and can actually output 120dB will feel something that resembles real life, the same kick when they press the throttle mid corner, instead of the normalised garbage we've had since 1993.
Are you suggesting, normalization (in meaning by compression/limiting) is done by rf sound engine? I would like to see some one who outputted racing sim volume at 120dB level. Even peek one.
I hope that could be a removable option. I'm a bit of an audiophile and compression and normalization may work for the layman. But people who know what they are doing want that s__t gone.. Here, just listen to this 100% uncompressed amateur recording of the pit-lane at the Nurburgring. http://www.mediafire.com/?o3lg5kyikv9i9d6 This guys setup might be equipped for that sort of sustained volume.
I'm not sure you understood me correctly. Gold suggested that sound in rF is flat and it is due to rFactor it self. I disagree with that. Sound is flat due to how is recorded and how is set in .sfx files. There is nothing against to do it more dynamic. You can set on-coast engine sound very quiet, and on-power very loud. After all max volume out and test endurance of your ears. So it has nothing to do with rf's sound engine. Modders might change that, but I believe they intentionally don't because they don't want to tune it up for very high power soundsystems. Most of simracers will not turn volume out to 100dB average and 150dB peak because it doesn't make sense (they should use then ear protectors which makes sound level lower and dynamics more flat; they also have neighbours). With volume set to acceptable level (usually for loudest sounds) level of quiet sounds will be unacceptable.
Don't actually know any details yet except hearing mention of wall/tunnel reverb/echo and surround, etc.
If the sound volume is too high within rF itself it doesn't seem to compress but clip the wave (similar to guitar overdrive effect), at least on my machine. On the other hand I noticed some stereo panning weirdness happen, so maybe there is compression. It's noticeable in this video, when cars get side-by-side, the engine sound of camera car suddenly pans to the left. Clipping clearly occurs when the car drives by the fixed camera on the wall. PS: this is an old video, I was experimenting with extreme "// ATTENUATION" values.
Clipping sound wave in digital domain is natural effect if two or more waves are added together. It's due finite range of values representing the sound sample. For example for 16 bit, there is 65536 levels (between -35536 and 35535). If added to waves, it may occur that 2 values gives number greater than max available or lower than min available. It must be clipped - there is no another way. You may use limmiter but it will only avoid distortion smoothing high level peaks. In result you will get compressed (gain compressed) signal, which definitely has lower dynamic (volume difference between low volume sounds and high ones - because the last ones has been limited) So, I don't know what you want to say referring to sound clipping. It is not related to rF and cannot be improved in other way than limiting signal. but it cause opposite effect to what Gold asked for. BTW: Another way to avoid clipping (and stay with maximum possibly dynamic) is to turn down the volume of source samples (in sfx files or in recordings). Again, it's up to modders, not to sound engine. BTW: seeing a future , please do not reffer to 24bit (or greater) recordings. It changes nothing. It makes only better resolution (quality: less aliasing, less quantisation noise) of sound. 24bit maximum volume wave (range -8388608 to 8388607) has the same played volume level as the same wave recorded at 16bit. So added 2 loud 24bit samples will also cause sound clipping. I suggest to read about sound processing in digital domain. Look out for subjects like: mixing, gain compression, limiting, harmonic distortion PS. Overdriven sound on this video may came from too high recording volume, not too loud sounds in rF. Just too high value in windows mixer (what you hear/stereo mix recording input)
As I said that was due to extreme values in ATTENUATION section of the sfx. I tried to achieve much greater difference in volume with distance. It worked, but had nasty side effects when using onboard cameras. This is one of the reasons why I'd like rF2 to have a completely separate mix for onboards in sfx or at least an adjustable parameter for ratio: own car cockpit sounds vs. other cars+ambient.
Yes it should have. Maybe there just a single parameter to set in-car volume ratio would be enough However if you end up with oversteering, just turn down a volume of played sounds I worked hard on sounds for our mods and had no problem with ovesteering. But in result external sounds are significantly quieter than internal one which is noticeable on external cameras. But I believe a noise inside a racing is very loud so other car sounds should be dumped more. PS. I did XLS sheet to make sfx files for whole mod. I makes me able to control all souns separately as well as globaly as well as single sections. I could share it if you want.
I'm too old for this 'quick reply', it lost my damn post! :S I agree with people who said you need more dynamic range. Realistic levels aren't a good idea, as I don't like bleeding ears, but most games and mods have so little dynamic range, that is the other extreme. However, the rf1 sound engine isn't incapable of this. The problem is, logically, the engine has to mix lots of sound (road/wind/tire/engine/gear/opponents/crowd..), but it does not amplify them into clipping / distortion. So if more sounds are added and the volume is already 100%, then you get some sort of compression going. Mods often use sfx volume multipliers, trying to make sounds louder, but the sum of all sounds can never be >100% I believe if you use samples that are near 0db, so 'max volume' , and use volume multipliers greater than 1 in the SFX, *AND* you have multiple noises like wind/road/engine, things get compressed. I would put money on it being very possible in rF1 to get like a 30dB dynamic range from idle to 200mph, which would work quite well with good headphones for example. The trick is to either use soft samples, or better still, use very small volume multipliers in the SFX file. Theoretically, I think *really not 100% sure!* the following is true ish: A) 10 samples at 0db (max volume) each having a volume multiplier of 0.1 in the SFX will result in linear adding / loudness of sound, the total can become 1, i.e. 100%, without compression. B) 10 samples at 0db (max volume) each having a volume multiplier of 0.25 in the sfx will result in linear adding of volume up and until 4 samples are playing, but when more samples start playing, compression begins. C) certain 90s F1 mods with lots of samples at 10x sound multiplier are silly, you still hear about the same noise even when you turn the engine sound down to 10% in game, because 10% of a 10x multiplier is still 1x i.e. 'normal' .. EDIT: of course if you ever do a mod with 30dB dynamic range, you'll only have to wait 1 minute before people say that the sound isn't loud enough, as if they can't find the volume knob on their speakers?? sigh
This was 5 minutes work, not a HUGE dynamic range, but given 'home user conditions' .. Idle, 30mph, full blast to 200mph, hard brake to idle, short blip. Mp3 in the zip file!
I'd still like the option to have correct levels should you want them. If you went all out for a $20,000 motion sim wouldn't you like to have the option to wear hearing protection just like a real driver. I think the biggest flaw is startup and idle sounds. Full throttle is obviously severely limited. Idle should be like sitting in a race car. Loud and low. Very little bass makes it through with the current schematic. All this shouldn't be too hard a thing to calibrate with reality. All you'd need is a decibel meter and a loud car. Again this isn't for everyone and level compression should be implemented just with the ability to shut it off.
Are you suggest, that sounds originaly should be prepared with high dynamics, and then compressed by a game to satisfy audiosystems of most players (with option to disable it for a few)? Don't you think that would be easier, that a few men on planet who want to race with hi-dynamic sound, will modify sfx files to match your needs? After all there may be another problem - for modders. Should they prepare sounds for cars referring to hi-dynamic specification or to home sound systems. Do you think that modders (sound creators) are able to test sound with such high loudness? I can bet, finally most (if not all) mods will be again prepared for compressed output. It would be worse than today, because of impossible to find correct volume by listening compressed output.
Yes. It would only need to be clear so that amplification on the final end doesn't introduce distortion or hum. Maybe as a whole the way sounds are produced should be changed. Instead of taking a sample the sound of a single combustion cycle itself should be produced and repeated for the number cylinders and RPM. That would make a misfire sound more realistic. etc. Kind of like how Warthog handled engine power in Richard Burns Rally. http://img.photobucket.com/albums/v504/F12Bwth2/4.jpg
Clear samples are not enough in that case. I referred to problems with volume settings while hearing compressed sound. For example for some settings of compressor/limiter, you may play source with different loudness and hear no difference. Only way is to work with samples without compression. But how to do that if it requires turning volume up to 120dB? Note also, that compression destroy original characteristic of recording. But sound in RBR is not synthesized from single explosion samples. Curenlty the best way is to using samples. First of all, sound recording is easier process than finding correct values for synthesizer. Second, it's almost impossible to get sound of real various cars using a synthesized methods. Maybe using some kind of backward transformation (something like sound reverse engineering). But guess, it is kinda high science .
I agree with this 110%. That road noise of an oncoming or outgoing car is missing in rFactor. It really should be there. However, you have a fine line to work with here. What if 10 cars were all coming down the straight at the same time (or even 20-30-50+) and rFactor decided to produce all of those road noises at once? That could be an issue. So that's something that would have to be dealt with. I do think it's possible, though.
This discussion is too technical for me. All I see is a possible paradigm shift, like gaming steering wheels were back in 1995 and FFB was back in 2001. When a real racing car starts up next to me in a garage it scares the SHYIT out of me. This does not happen in any game. Headphones or 150+db soundsystem, the option should be there to output non normalised sound (rF or Mod induced). If you use lower frequencies to add to the punch of lower volume sounds (eg idle) you can recreate reality. It doesn't need to make your ears bleed, it just needs to sound dynamically recognisable. To my amateur ears what sounds unrealistic fundamentally are the use of bass and the use of echoing/echos/reverberation. A simulated racing car does not need to output 120db to sound impressive. Aslong as the sound it makes can be contrasted to ambient sounds (wind, crowds, etc) you can feel its power in the reverberation.
It just because you don't understand what dynamic means and what makes you scary in your example Dynamic means great difference between sound volume. It makes you scary because the difference makes air pressure kick your ass. It is very simply but also impossible until you will use very high volumes While developing sound for game you have 2 options 1. make sound dynamic - players who have rigs in a garage or basement or so, will be able to turn volume up to 150dB and be satisfied (or end up in hospital, whatever) - players which has computers at home - will turn volume down, which will make most of sounds in game mute. Try to play example made by Niels with standard volume, you will get what I mean. 2. make sound more flat, playable on most player's rigs So again, you put your request in rF2 thread, in Sound Info one, suggesting that the issue depends on sound engine. No, it is not true. It depends on how creators prepare sound layer for cars. What do you think, which option they will chose, 1 or 2? And there is also technical part of issue. Setting volume of single samples is not enough. Real racing car sound kicks constantly, without changing revs or coast to power. Unfortunately recordings it self, are mostly flat, because made from YT videos, or with using lo quality equipment. Recording sound with such dynamic if a very hard work and expensive also. Just note that sound with 120-150dB is hard to record with quality without use of special equipment. More over, 16bit recordings (mostly used) has maximum theoretical dynamics around 96dB (mostly lower). It is impossible to recreate nature of real racing car sound that way. And finally, which has been said a lot of time - to get what you need, you have to play results with really high volume. But no one will try it at home. Because it will brake glass in windows.
