rFactor 2 engine and car sound. Room for improvement

I am not sure for others. For me, the rF2 engine sound is not that good in comparison with real ones, and need to improve. I made a video on NSX sound.

"Engine sounds of many different Honda NSX versions were recorded by PhuShizzle.
I played out his video by using a Spectrum Analyzer directly connected to soundcard outputs to monitor the sound.

The top chart on the screen is FFT spectrum with x-axis in logarithmic scale. As you can see and hear that how rich of real engines sound with rich lower notes below 200Hz. The spectrum is clean and with unique signature for each version of cars, which can be clearly heard.

By comparison, rFactor 2 NSX sound is quite different from real ones’: Lack of all low notes below 200Hz; Lost of inner details of engine sound and almost no matched signature of real cars’ sound. I increased sound level for frequencies below 200MHz by 12 dB, using an equalizer. But still cannot make it sounds close to real ones.

You may need a set of good headphone to listen if you do not have a good amp and speakers."

 

Oh god, don't get me started on this. Driving the FR3.5 which sounds like it's idling or almost off at anything BUT 100% throttle, even at 5000 RPM. Oh god, don't get me started with sound. Ok, I'm going to stop now.

 

I would say 90% of the sound impression depends on quality of sound samples used. I would guess NSX as a modded car from rF1 doesn't use very high quality sound samples, as far as I know the sounds were just taken from rF1 version.

 

"Samples"? the "Samples" showed here in spectrum distribution are totally diff from real ones. It seems they took wrong sound base for the NSX, or they recorded a wrong engine sound. Missing lower sound modes below 200Hz, though. Or they recorded the sound at total enclusured cockpit with cutting all freq < 200Hz.

 

There is always room for improvement. To be fair though the NSX is probably one of the worst sounding cars. The tire squeal in that car sounds odd too.

 

It squeals to early. It annoys as it is, and isn't realastic. Even the most bad street tires don't do that. I'm just hearing üüüüühhhhhhhhhhhhhhhhhhhhhhyyyyyyyyyyyyyyyyiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiii when driving that car.

 

Not sure on the tire squeal...
For engine sound, I tested the Corvette GT and Nissan GT-R GT, they are just bad as the NSX. It seems to me that really come down to how they record the engine sound. I do not think they did right.....

 

AFAIK ISI doesn't record the sounds, Greg Hill does.


http://www.soundwaveconcepts.com.au/

 

Thanks for the link. Pretty disappoint ..... on the rF2 engine sound work.

 

It is well beyond my scope to speak of the engine being poor, but I greatly agree the sound reproduction could be better. From the actual reproduction, to the lower tones, to the volume and range (distance) which the sound is reproduced. I remember walking to Monza and from a couple of km I could hear the GP2 cars. When I went to Spa, I could hear the F1 engines much fiercer and louder while driving and then walking to the circuit. Only R3E has some form of representation of the last part.

 

It's not just recordings. It's also about what the audio engine does with those recordings, how it works and "plays" with those recordings to dynamically change the sound characteristics to make things seem complex, "deep" and real, rather than simple and digitized. There is A LOT to good sound like volume, pitch, tone, reverb, delay, reflections, fade, echo, rolloff, decay, attenuation, shape of attenuation, distortion, specialization, filters (eg. high and low pass), and on and on and on.

I agree though, sound recordings/samples are obviously extremely important as well.

There is an old rF1 video that the company Natural Point uses to market TrackIR. I think it uses a Formula Nippon mod. Anyways, there is hardly any dynamic range in the video, which is obviously not how it should be. There is also a lot of un-clear and distorting samples, which is also obviously not how it shouldn't be, HOWEVER, the car ALWAYS has a powerful sound - not only at 100% full-throttle, but at part throttle and even completely off-throttle, and even at off-throttle and low revs at the same time. This is what real cars are like, always a sense of audio power even at off-throttle and low revs.

Please don't judge the below video's sounds as whole, I know it hardly has any dynamic range, and also the sounds are distorted to hell, so I know it has issues, all I'm saying is this car sounds "full", "thick", powerful, and "loud" even at low revs, low-throttle, and off-throttle, which can't be said about 95% of cars in sims. I mean, listen to how rough it sounds even on the blips, and even after the blips as he's entering the corner with no throttle on the car (or very little throttle), it always sounds like a beast, rather than only sounding like a beast when at 100% throttle like 95% of cars do.

I think I know why the video sounds like it does. Even the sounds of the curbs distort, hahahahha. I think there was a problem with some audio settings when the person made the video and everything's volume was raised way too high, regardless, I still stand by what I said about how "thick", "full", "loud", and powerful it sounds even when at off-throttle and low-revs. No, I wouldn't want to play with these sounds like this, the distortion would kill me, lol, I'm just trying to send-out the idea that the off-throttle and part-throttle sounds also need to sound powerful, whereas 95% of cars, default and third party mods, sound like junk UNLESS you are giving the car full throttle - then, and only then, can any sort of sense of depth and power be heard and "felt".

This can even affect your driving (when the engine sounds like it has 5 hp unless you bury your foot to 100%, full-throttle).

 

I try to be "objective" on this. There is no "subjectivity" or "speculation" on this...
In my video, the Spectrum Analyzer connected to sound card to monitor both: 1) real car engine sound recording from PhuShizzle video; 2) rF2 sound output from rF2 real-time playing. Side by side to compare...
The results as you can see that rF2 engine sound is totally diff from real ones, as I described in the video description.

 

The sound engine is quite capable (Miles), but it depends on the recording samples & subsequent sound file mixing. For an example of how good the results can be, try the DRM Revival mod in rF1.

 

You should make that comparison with some ISI car then. NSX is not ISI-made, it says on website: "This was a collaboration with Siim Annuk and Niels Heusinkveld, who developed this car originally as a mod for rFactor 1." Those guys probably had no access to any real engine recordings, 99% of rF mods have more or less fantasy sounds.

 

How you do things, and the quality of material resources, is the crux of the matter here, because the sound engine in RF2 is, at this point, as good as it gets for a racing-sim (IMHO).

I think the issue with RF2 (and most commercial sim titles, IMO with few exceptions) is somewhat poor character samples (lacking "texture" and depth), unoptimized volumes (lacks dynamics a bit) and, as mentioned, lacking accuracy in many, many cases.
Just by looking at some of the .SFX files parameters and the .WAV sound samples, I wouldn't do things that way if it was me doing them. But then many of us look at things (or listen/feel) in different ways and can think of completely different things and solutions all the time, in everything.

The current version of Miles Sound Engine, which RF2 is using, does lack effects (reverb, echo, low/high pass and a few other effects we would all like to see/hear implemented) but, I suspect, even some of those effects could be just as distant as a basic update for the sound-engine package (ISI may wish to enquire Miles for it?).
I'm mentioning this by looking/listening to, I believe, what Sector3 (ex-Simbin) did with R3E, which also uses Miles (just like RF2) but does seem to have the mentioned features/effects in the sound engine.

The problem -and respective justification- with sounds for commercial games is that, most times, there's a budget and, most often, also a time restraint (to reach a target, for example, to time of release), so some things must be compromised. That's why most cars in games don't sound as good, nor accurate, as many of us wish (sometimes even to the author).
So, for instances, (re)using sound samples from a copyrighted/proprietary library, which may or may not be fitting the particular car(s) and situation(s), does happen. Because it cuts time (it's easier), because it's (supposedly) good enough for the objective.

The case of the NSX is a bit special and, as mentioned, it is not accurate regarding sound, as even Niels confirmed (see here).
The sound it uses came from the Austin Healey 3000 of GT-Legends (see video to compare audio). Niels used its sounds as a base (inside and outside samples mixed, loops retouched too, I think) as he thought it sounded close enough.
I'm not sure if the samples of the GTL Healey were originated from a recording of a Jaguar straight-6 XK engine (D-Type, E-Type) but it sounds a lot like it.
Being a specialty of mine p) I offered to assist him with all new sounds from recordings of the real counterparts but I got very busy with real life stuff in that period and, unfortunately, couldn't help in time for the release of the RF1 car.

 

Just for fun I'd like the OP to try this with the Howston. From the outside that car sounds pretty good IMO.

 

Totally agree, sounds can affect not only immersion but a lot your driving.

As we discussed in another thread you can think physics or setup is incorrect due to incorrect relative levels of sounds (low revs, skidding).

 

The relative sound amplitude levels for different RPMs is mod-specific setting. For example FISI has:

VS_INSIDE_POWER_ENGINE_1=FISIR\fisir_in_onidle.wav
VS_INSIDE_POWER_ENGINE_2=FISIR\fisir_in_onverylow_1.wav
VS_INSIDE_POWER_ENGINE_3=FISIR\fisir_in_onverylow_2.wav
VS_INSIDE_POWER_ENGINE_4=1.10,FISIR\fisir_in_onlow.wav
VS_INSIDE_POWER_ENGINE_5=1.20,FISIR\fisir_in_onmid.wav
VS_INSIDE_POWER_ENGINE_6=1.30,FISIR\fisir_in_onhigh.wav

You can see coefficients 1,10 1,20 1,30 for last three sounds, this is the boost factor. So high rpm has more amplitude boost set in this mod as well.

 

OK, here is the video for Corvette C6R GT2 (rFactor 2 vs real) for both in-cockpit and outside of the car.
You can see and hear though to compare. How/what do you think? (note: the in-cockpit real engine sound waveforms are clipped....distorted). The rF2 Corvette's sound far better than NSX to me. You may need a set of good headphone to listen if you do not have a good amp and speakers. For inside-cockpit, the rF2 spectrum seems very similar to that of real car engine sound. However, the real sound spectrum showed clean notes and with strong lower tones. The rF2 sound seems lack of details and ....muddy..
By listening and looking into carefully, I bet that ISI might have actually used the original recording for the Corvette, but mixed with too much env sound and effects (you can tell from both spectrum and waveform charts), and OVER processed, which smeared whole spectrum, and hence lost inner details and some signature lower notes of the original engine sound. I wish they could offer an option allowing user to select between clean "real/original" sound and "gaming sound" (like this one). I am pretty sure they could do it since they used original recording.

 

speed, could you repeat that one please, I didn't quite get it. (just pulling your chain)