GT5 Sound Thread
- Thread starter Marry_Me_GT
- 5,638 comments
- 545,730 views
You obviously dont have a passion for cars and their unique and individual exhaust note that each car produces. For myself, I wont race a car that sounds like crap. Hearing proper sounds are crucial to racing as they give you reference points as to what rpm your at and when you need to shift. If you learn the sounds of a car you never need to look at your tach or speedo but in all GT games atleast for me it's really hard to tell.
- 738
- RaLpH89 / GTP_RaLpH89
IMO their sample in low RPM is kinda spot on, like the audio R8 at 1k to 4.5k RPM in GT5p, but when you rev it all the way up to red line it's clear that they are missing some samples. Engine stress and other missing samples that they didn't applied. Has anyone notice that almost all the GT5 game we've seen and played so far are all compressed data. 
. They haven't even showed what they've been cooking in the past 5 years, except for some screens and videos. Other than that it's all classified. Come on PD this is not the damn military!
Edit
Even GT5 prologue is missing a lot of cars and features!
. They haven't even showed what they've been cooking in the past 5 years, except for some screens and videos. Other than that it's all classified. Come on PD this is not the damn military!Edit
Even GT5 prologue is missing a lot of cars and features!
- 1,945
- Treviglio
- PAPPACLART
- PAPPALCART
Isn't that the point with a simulator, the idea beint to simulate driving a real car? Surely we can expect comparisons to be made?
The cars look pretty real to life, they handle pretty real to life but why can't they sound real to life? I would think capturing the sound would be easier than the 2 former.
- 4,878
- CoolColJ, GTP_CoolColJ
So people rag on the sound of GT5, or what they have heard of it so far. But one thing PD gets right is the doppler effect and cars have good external audio. PD are not going for the bombastic sounds, but more true to life feel
LFA is the car I'm most forward to in GT5!!
There are some that champion NFS Shift's sounds, well they are definitely not accurate, over processed and distorted.
LFA has a glorious F1 style wail on the outside. Shift's LFA doesn't even remotely sound like that, hell even the LFA doesn't sound like that in car.... Just like I thought Hollywood car sounds...
The same crappy sound in car and outside!
----------------------------
the real LFA
and in car
LFA is the car I'm most forward to in GT5!!
There are some that champion NFS Shift's sounds, well they are definitely not accurate, over processed and distorted.
LFA has a glorious F1 style wail on the outside. Shift's LFA doesn't even remotely sound like that, hell even the LFA doesn't sound like that in car.... Just like I thought Hollywood car sounds...
The same crappy sound in car and outside!
----------------------------
the real LFA
and in car
Last edited:
- 370
- West Yorkshire
- DRambo
- 10,403
- GTP_Spag69
- spagetti69
I can confirm gt5 will have sound.
- 1,806
- Belgrade
- FrancSRB
This sound pretty good!!!!
http://www.videogameszone.de/aid,70...cus-04-so-realistisch-klingt-der-Sound/Video/
http://www.videogameszone.de/aid,70...cus-04-so-realistisch-klingt-der-Sound/Video/
- 846
- GTP_juniordee3
- Oddbug
- 192
- Finland
No way! Awesome! 👍
- 8,725
BIIIG post, sorry guys... I'd appreciate it if you read it though! 
I'm fed up with uniformed people crying about engine sounds. If you like NFS:Shift's engine sounds, play NFS:Shift. Personally, I think it sucks (in many ways) - sure it sounds exciting, but nothing like a real car doing anything!
As for PGR4, I was very impressed with the quality of the samples used for that Tamora, but remember there were only 130 or so cars in PGR4... GT5 will have over 1000. Does anybody have any idea how much data that corresponds to?
Per car:
3 samples across the rev-range (conservative?)
x 3 for each "quality" - off-load, mid-load and full-load.
Plus one idle...
10 samples per car, > 10 000 overall.
Say a typical sample is in the region of 2 seconds long (minimum, for effective looping), for 350 minutes of samples all-told. Thankfully we can save a bit of space because the sounds will all be mono and positioned by the audio hardware for spatial effects etc.
Mono CD quality is 1411200 / 2 bps, so ~700 kbps, and 1660 MiB for all of the samples.
Triple this for separate engine / intake, interior and exhaust sounds and it's becoming a substantial chunk of the available BD space - especially when you consider that a good loop is probably > 4 sec in duration (double it) and you'd want to subdivide the rev-range into at least 4 chunks, not 3 (add 30%)! Sure, you could use a lossy format such as Vorbis or MP3, but it's still a big chunk of data.
Moving on... You could have the best quality samples in the world, but you still need to implement them properly. PGR4 sounds uncanny because the way the engine responds during gear changes is really quite strange - GT5 has excellent downshifts and, if you use a manual clutch, superb upshifts!
And comparing real-life recordings to game sounds is pointless, because you cannot accurately reproduce the soundstage. Take that vid of the Zonda, there is a huge amount of (very complex) reverb coming off the buildings, meaning you hear the exhaust note from the reverb before the car has passed you. Also, its exhaust is quite resonant and the frequency mix changes over a very small rev range. I have never heard a good representation of a Zonda in a video game.
The F1 in GT5
demonstrates that PD are using environmental reverb, as you can hear the exhaust note coming off the surroundings before the car is anywhere near the camera (listener).
Another thing to bear in mind is dynamic range. Our ears / brains have tremendous dynamic range. Audio formats generally do not. If you try to increase the dynamic range, you lose sensitivity. So the sounds need to be rendered at max detail (requires excellent recordings, using all of the bit-space with no clipping) and then processed by the audio engine in the game. PD do this. Their handling of dynamics is superb, except that the "aperture" does not change.
Our ears adjust according to the loudest sound audible. As far as I can tell, GT5 normalises according to the loudest sound that may be produced (I.e. it has to be turned up LOUD to appreciate it!). Think how loud a ticking clock sounds in a quiet, empty room, but is barely audible over quiet conversation...
As for synthesis, as I mentioned on a similar thread, I too believe that PD uses a synth / sample hybrid approach.
I myself am working on an engine sound synthesiser and the results are similar to the implementation in LFS, except I haven't bothered to develop the complex filters for the exhaust piping just yet. This is because, as in LFS, I am having trouble finding a suitable pulse sound. Once this is achieved, I envisage that synthesis will be far superior to any sampling technique. The beauty of an internal combustion engine is that its sound is produced crudely and is formed through simple geometries (outside the cylinder head), so we can achieve much higher fidelity than with a musical instrument.
Having said all this bumph, I am incredibly impressed with the sound effects PD had developed for GT5. GT5:TT was superior still. Little details such as the distortion caused by gas flow exiting the exhaust pipe, and the way it phases as the car drives past, are beautifully recreated. 👍
Now I don't claim to know very much about audio in general, much less about digital representations / reproductions of audio, but my advice to all is to listen to GT5 in a quiet, non-reverby room with a half-decent 5.1 setup (mine was £30...) and listen to it LOUD and with the audio settings set to "large theatre".
The latter opens up the dynamic range so the subtleties can be heard, if you don't believe it's even half decent, you're beyond help!
I'm fed up with uniformed people crying about engine sounds. If you like NFS:Shift's engine sounds, play NFS:Shift. Personally, I think it sucks (in many ways) - sure it sounds exciting, but nothing like a real car doing anything!
As for PGR4, I was very impressed with the quality of the samples used for that Tamora, but remember there were only 130 or so cars in PGR4... GT5 will have over 1000. Does anybody have any idea how much data that corresponds to?
Per car:
3 samples across the rev-range (conservative?)
x 3 for each "quality" - off-load, mid-load and full-load.
Plus one idle...
10 samples per car, > 10 000 overall.
Say a typical sample is in the region of 2 seconds long (minimum, for effective looping), for 350 minutes of samples all-told. Thankfully we can save a bit of space because the sounds will all be mono and positioned by the audio hardware for spatial effects etc.
Mono CD quality is 1411200 / 2 bps, so ~700 kbps, and 1660 MiB for all of the samples.
Triple this for separate engine / intake, interior and exhaust sounds and it's becoming a substantial chunk of the available BD space - especially when you consider that a good loop is probably > 4 sec in duration (double it) and you'd want to subdivide the rev-range into at least 4 chunks, not 3 (add 30%)! Sure, you could use a lossy format such as Vorbis or MP3, but it's still a big chunk of data.
Moving on... You could have the best quality samples in the world, but you still need to implement them properly. PGR4 sounds uncanny because the way the engine responds during gear changes is really quite strange - GT5 has excellent downshifts and, if you use a manual clutch, superb upshifts!
And comparing real-life recordings to game sounds is pointless, because you cannot accurately reproduce the soundstage. Take that vid of the Zonda, there is a huge amount of (very complex) reverb coming off the buildings, meaning you hear the exhaust note from the reverb before the car has passed you. Also, its exhaust is quite resonant and the frequency mix changes over a very small rev range. I have never heard a good representation of a Zonda in a video game.
The F1 in GT5
demonstrates that PD are using environmental reverb, as you can hear the exhaust note coming off the surroundings before the car is anywhere near the camera (listener).Another thing to bear in mind is dynamic range. Our ears / brains have tremendous dynamic range. Audio formats generally do not. If you try to increase the dynamic range, you lose sensitivity. So the sounds need to be rendered at max detail (requires excellent recordings, using all of the bit-space with no clipping) and then processed by the audio engine in the game. PD do this. Their handling of dynamics is superb, except that the "aperture" does not change.
Our ears adjust according to the loudest sound audible. As far as I can tell, GT5
normalises according to the loudest sound that may be produced (I.e. it has to be turned up LOUD to appreciate it!). Think how loud a ticking clock sounds in a quiet, empty room, but is barely audible over quiet conversation...As for synthesis, as I mentioned on a similar thread, I too believe that PD uses a synth / sample hybrid approach.
I myself am working on an engine sound synthesiser and the results are similar to the implementation in LFS, except I haven't bothered to develop the complex filters for the exhaust piping just yet. This is because, as in LFS, I am having trouble finding a suitable pulse sound. Once this is achieved, I envisage that synthesis will be far superior to any sampling technique. The beauty of an internal combustion engine is that its sound is produced crudely and is formed through simple geometries (outside the cylinder head), so we can achieve much higher fidelity than with a musical instrument.
Having said all this bumph, I am incredibly impressed with the sound effects PD had developed for GT5
. GT5:TT was superior still. Little details such as the distortion caused by gas flow exiting the exhaust pipe, and the way it phases as the car drives past, are beautifully recreated. 👍Now I don't claim to know very much about audio in general, much less about digital representations / reproductions of audio, but my advice to all is to listen to GT5
in a quiet, non-reverby room with a half-decent 5.1 setup (mine was £30...) and listen to it LOUD and with the audio settings set to "large theatre".The latter opens up the dynamic range so the subtleties can be heard, if you don't believe it's even half decent, you're beyond help!
CodeRedR51
Premium
- 55,344
- United States
^^^
I say we end the thread with that post. 👍
I say we end the thread with that post.
👍A lot of explanations proving actually...nothing. GT sounds still suck hard and no matter how much you explain about samples, Hz, reverberation they still suck.
If you can't do it, you don't do it. Not doing it right a 1000 times does not mean you're doing it right.
I don't need to start giving real and virtual examples again to prove the sounds are very unreal, right?
If you can't do it, you don't do it. Not doing it right a 1000 times does not mean you're doing it right.
I don't need to start giving real and virtual examples again to prove the sounds are very unreal, right?
- 8,725
I wasn't trying to prove anything. I'm trying to get people to open their ears! When was the last time you heard a car like the ones in NFS Shift?!
if you don't believe it's even half decent, you're beyond help!
...
If you can successfully stand up to that challenge, let's hear 'em! 👍
EDIT: By the way, I appreciate there's potentially a language issue here...
Last edited:
- 1,945
- Treviglio
- PAPPACLART
- PAPPALCART
BIIIG post, sorry guys... I'd appreciate it if you read it though!
I'm fed up with uniformed people crying about engine sounds. If you like NFS:Shift's engine sounds, play NFS:Shift. Personally, I think it sucks (in many ways) - sure it sounds exciting, but nothing like a real car doing anything!
As for PGR4, I was very impressed with the quality of the samples used for that Tamora, but remember there were only 130 or so cars in PGR4... GT5 will have over 1000. Does anybody have any idea how much data that corresponds to?
Per car:
3 samples across the rev-range (conservative?)
x 3 for each "quality" - off-load, mid-load and full-load.
Plus one idle...
10 samples per car, > 10 000 overall.
Say a typical sample is in the region of 2 seconds long (minimum, for effective looping), for 350 minutes of samples all-told. Thankfully we can save a bit of space because the sounds will all be mono and positioned by the audio hardware for spatial effects etc.
Mono CD quality is 1411200 / 2 bps, so ~700 kbps, and 1660 MiB for all of the samples.
Triple this for separate engine / intake, interior and exhaust sounds and it's becoming a substantial chunk of the available BD space - especially when you consider that a good loop is probably > 4 sec in duration (double it) and you'd want to subdivide the rev-range into at least 4 chunks, not 3 (add 30%)! Sure, you could use a lossy format such as Vorbis or MP3, but it's still a big chunk of data.
Moving on... You could have the best quality samples in the world, but you still need to implement them properly. PGR4 sounds uncanny because the way the engine responds during gear changes is really quite strange - GT5 has excellent downshifts and, if you use a manual clutch, superb upshifts!
And comparing real-life recordings to game sounds is pointless, because you cannot accurately reproduce the soundstage. Take that vid of the Zonda, there is a huge amount of (very complex) reverb coming off the buildings, meaning you hear the exhaust note from the reverb before the car has passed you. Also, its exhaust is quite resonant and the frequency mix changes over a very small rev range. I have never heard a good representation of a Zonda in a video game.
The F1 in GT5
Another thing to bear in mind is dynamic range. Our ears / brains have tremendous dynamic range. Audio formats generally do not. If you try to increase the dynamic range, you lose sensitivity. So the sounds need to be rendered at max detail (requires excellent recordings, using all of the bit-space with no clipping) and then processed by the audio engine in the game. PD do this. Their handling of dynamics is superb, except that the "aperture" does not change.
Our ears adjust according to the loudest sound audible. As far as I can tell, GT5
As for synthesis, as I mentioned on a similar thread, I too believe that PD uses a synth / sample hybrid approach.
I myself am working on an engine sound synthesiser and the results are similar to the implementation in LFS, except I haven't bothered to develop the complex filters for the exhaust piping just yet. This is because, as in LFS, I am having trouble finding a suitable pulse sound. Once this is achieved, I envisage that synthesis will be far superior to any sampling technique. The beauty of an internal combustion engine is that its sound is produced crudely and is formed through simple geometries (outside the cylinder head), so we can achieve much higher fidelity than with a musical instrument.
Having said all this bumph, I am incredibly impressed with the sound effects PD had developed for GT5
Now I don't claim to know very much about audio in general, much less about digital representations / reproductions of audio, but my advice to all is to listen to GT5
The latter opens up the dynamic range so the subtleties can be heard, if you don't believe it's even half decent, you're beyond help!
Thats a very good write up, but unfortunately GT5P does not sound very good in most cases. Sure there are exceptions but few and far between. As for sampling sounds, well I am sure a 50GB disk, or even 2 for that matter should leave plenty of room for some quality sound recordings. With Lossy codecs now, quality is more or less kept but at a substantially smaller size.
I think most people are mainly concenred how the supercars or more sportier cars sound, as most 4 pot eco cars all sound the same, so I don't think it is really important to sample all of those eco boxes which amount to a large portion of the cars in the game.
As for surround sound, well I have an Onkyo Tx-sr507, with 5 energy speaker rated at 100 watts at 8ohms with a 200 watt sub. I have cranked it up loud, but still although sound does come from all 5 speakers, the sound field is very narrow, and is almost mono like, there is no depth, it is awful!!.
- 8,725
Firstly, thanks for the kind words, but I just cobbled it together!
Anyway, I agree that the result of the engine sounds in GT5
is a bit lacking in most cases, but I believe this to be due to bad mixing of the different components of the vehicle sounds. The time trial was much better in this regard. I imagine GT5 proper ought to be similarly artistically optimised. The "Tuned cars" in GT5 all (well, almost all) sound fantastic, the dynamic range really playing its part there.I agree about the 50 GiB (100 for dual layer!
) potentially leaving enough space, but I forgot to factor in the modified sounds, many of which will of course be shared... also, I have no idea how much space car models / tracks take up - they must be huge!True enough also, although I've been disappointed by many (unmodified) super cars in "real life", as they can sound quite whizzy compared to full-on race machines - often themselves sounding a bit 🤬. Of course GT represents the full gamut, hence so-called "econoboxes" are always welcome, and really don't sound so similar as you might at first think, especially when unmodified. It's a bit like saying GMs sound like Fords to the V8 brigade - the differences are subtle, but that's what counts.
That is odd. Mine are only a set of Creative, err, SBS580 and I get great depth and envelopment. I'm using the optical output in standard 5-Ch to an Auzentech X-Fi prelude in my desktop PC, so the soundcard is doing the "decoding" directly.
Of course, most of the time (in replays at least) the camera is facing the origin of the sound, so the additional speakers are merely catching the reverb, which is really rather subtle.
Look, we are comparing one game's sounds to another game's sounds. Pages back there was a consensus many games produce better sounds. We don't need to blow our ears of or buy new stereo systems when we can hear other games perform substantially better. If it's better, it's better. All fanboys try to change the way things work with the sole purpose of proving GT as "the best".
A guy showed a TSU mod he had made and it sounded heaps better than anything in GT5.
Talking about acoustics and sounds bouncing of walls is also bullcrap, because I can find you tons of supercar videos shot in the field where there's nothing to reflect the sounds, and the cars still sound meaner than anything.
And last but not least, off course we are going to compare GT5 to reality, especially since many games have gotten really close to reality. GT5 is the "real driving simulator" right? So at least the sounds should be real.
A guy showed a TSU mod he had made and it sounded heaps better than anything in GT5
.Talking about acoustics and sounds bouncing of walls is also bullcrap, because I can find you tons of supercar videos shot in the field where there's nothing to reflect the sounds, and the cars still sound meaner than anything.
And last but not least, off course we are going to compare GT5 to reality, especially since many games have gotten really close to reality. GT5 is the "real driving simulator" right? So at least the sounds should be real.
- 8,725
This is the cause of our disagreement. I don't believe a single game sounds anything like real life. Yes I've been dazzled by gravelly exhaust notes with a deep, pumped up bass backing, and swishy-fizzy noises lavished over the top - but it just doesn't sound anything like the real deal. I really dislike the rough sample blending, as in the GTR games, it just spoils the atmosphere.
I never said GT5 was the best, I think their audio engine is one of the best in a racing "sim", assuming I've correctly inferred what's going on. I reckon you'd be amazed at how important reverb is - in (some) old western films, gunshots sounded the same in the saloon as they did out on the prairie, as they did in the mountains - utter tripe!
And there was no such thing even approaching a consensus reached anywhere in this thread or others like it in the past.
I'm just fed up with Hollywood SFX being regarded as the ideal. If you want that, great, go watch 2 Fast 2 Furious or play NFS. If you want GT5 to sound like that, it's never going to happen, as PD seem to know how cars sound in real life, and they're working towards that - GT5
isn't quite there, GT5:TT was closer - GT5? I doubt it'll be perfect, but there certainly won't be enough whizz-pop-bang for the I've just played GT5
with some large over-ear headphones and the difference is insane! The spatial separation of sounds and the environmental reverb really come across, although London could do with more 'verb! One thing I've noticed is that "engine" noise is too heavily mixed (the lotus elise S1 sounds really tappety, as does the F40!!) and there is little-to-no intake roar - bit of a huge omission, really. 
Still, turbos sound nice!
- 1,660
- P3nT4gR4m666
One point I'd disagree with you on - I don't think you'd need quite the sample length you do for accurate engine noise. You're recording something that does the same thing thousands of times a second. Given that turbos and dump valves and the like must be overlays the actual revving itself shouldn't take more than a couple of millisecs at reasonably high sample rate?
- 8,725
That's a very good point! It's interesting, really. Most car engines do not sound the same from one point in time to another, due to minute changes in the RPM, the interference of many many resonances etc. etc. so making recorded samples into loops is quite a task. Using a really short loop just makes the detail drop out (it sounds flat, bland etc.), because you're not getting all of those variations. Looking at rFactor, all samples are at least 2 seconds long, some are even as long as 6 seconds. I guess it's the magic number to avoid obvious looping at higher RPM.
A good example of obvious looping in GT5
is the Tuscan, there's a really annoying burbly loop that just grates on me. It's been like that since GT3...I could bosh up a quick example of different sample lengths with GPL (single sample for all RPM, engine loads etc. - at least, for now) if you're interested, although I'm not very good at making looped audio!
So it wouldn't be fair.You could always try it yourself with your favourite audio player set to repeat!
Something I've noticed about the rFactor samples (I really enjoy the semi-arcadey sound in rFactor, actually...) is that the intake noise seems like its just the exhaust noise with a bit of an eq. tweak... At least the sounds I heard, anyway. Cheap, but seemingly effective?!
EDIT: Gah, I'm a 'tard! It was interior / exterior, not intake / exhaust... Makes more sense, don't fancy blending 6 samples at once!
Last edited:
Similar threads
- Locked
