Foobar2000 upsampling

[temhem]

A few weeks ago, I took Fewtch's advise and started experimenting with upsampling in foobar2000. I have terratec 6fire 24/96. Fewtch, said things sounded better when you upsampled, I tried and I saw an immediate improvement.

What is the improvement? Well, right now, for example, I am listening to Hilliard Ensemble playing some Italian madrigals. It is a good clean recording with a range of vocals. The changes between the sampling rates 44100 and 48000 that I hear are like this:

1. at 48 sounds open up, and sound stage widens
2. Again at 48, the higher voices become richer and more liquid. And that digital (metallic) sound most complain about becomes less pronounced.

I hear no difference between 48 and 96, so I am quite puzzled as to what it may be that gives me this impression. I am quite sure that the effect is not placebo, although the effects are more noticable with some passages more than others.

It says on foobar2000 that upsampling will not increase sound quality, but resampling up to 48000 may avoid issues with some hardware. Well, I would think they are reffering to SB soundcard, and not Terratec. So I really don't know what is going on.

I wonder if others have tried this, and what their opinions are.

Cheers to all...

 

[amol]

How about trying out the ASIO plugin for foobar2000.
Or even better i liked was the ASIO plugin for winamp 2.X series. You can get it here, i really noticed an improvement using the ASIO plugin for winamp and resampling it to 48 khz even on my Sblive card

You can get it here..
ASIO PLUGIN

Selecting the similar sample rate from the ASIO control panel of the card and using the resampling from the plugin itself to the similar sampling rate would be the best setup i guess. I am keeping the sample rate to 48Khz from the ASIO control panel for sblive card, and also usign 48 khz resampling from the plugin itself. I notice some really good improvement. i hope this helps you

 

[Glassman]

Quote:

Originally posted by amol resampling it to 48 khz even on my Sblive card

your concluion is wrong, it is a must-have on Creative cards, while they resample internally and you must prevent them from doing it.. it should have no effect on cards that doesn't resample (all cards with Envy24 and such inside)..

but there is more to resampling than just preventing some crappy creative from doing it on its own.. it is clearly evident that from resampling to 96+kHz it sounds just more natural, less grainy, less digital..

if you hear a difference between 44.1 and 48 and not 48 and 96 it is most likely that the oscilator for the 48 multiples has less jitter than that for 44.1 multiples.. try using 88.2kHz resampling if it has the same effect as 48 and 96 or if it's the same as 44.1..

 

[pbirkett]

I normally use Foobar 2000 with 16-bit dithered Kernel Streaming and default 44KHz frequency for my music, on my Terratec EWX24/96.

I also decided to try the upsampling of the music as fewtch said.

First off, I tried the SSRC resampler at 96KHz with 24-bit non dithered Kernel Streaming output. The result was a blue screen of death.

So instead I tried it with Direct Sound v2 output instead, and it didnt crash, and it did indeed sound different. It sounded worse to me. There was no drop in quality, but it did impose a softness to the sound and lost a little bit of power and drive. I tried it for a few days before switching back to 16-bit dithered KS.

I have read quite a few reports of upsampling, and a lot of the time they say it makes the sound softer but more refined. It definetly seemed to do this in my case.

 

[ADS]

Quote:

Originally posted by pbirkett I normally use Foobar 2000 with 16-bit dithered Kernel Streaming and default 44KHz frequency for my music, on my Terratec EWX24/96. I also decided to try the upsampling of the music as fewtch said. First off, I tried the SSRC resampler at 96KHz with 24-bit non dithered Kernel Streaming output. The result was a blue screen of death. So instead I tried it with Direct Sound v2 output instead, and it didnt crash, and it did indeed sound different. It sounded worse to me. There was no drop in quality, but it did impose a softness to the sound and lost a little bit of power and drive. I tried it for a few days before switching back to 16-bit dithered KS. I have read quite a few reports of upsampling, and a lot of the time they say it makes the sound softer but more refined. It definetly seemed to do this in my case.

I've noticed the sound quality drop with Directsound 2.0 also, but I thought it was just me hearing things. Glad to hear I'm not the only one. I cannot tell a difference between any of the other output methods, but Directsound 2.0 results in a definate quality drop.

 

[Glassman]

yes this is it

you should use ASIO output plugin and 24bit dithered output for whatever you play.. why using 16bit output to 24bit card? we can do better so why not..

 

[pbirkett]

I just cant get ASIO to work at all on Foobar with my EWX24/96. Get the following...

"ERROR (foo_output_asio(dll)) : unsupported output data format."

Kernel Streaming still blue screens when using 24-bit 96KHz upsampling, but not when using 16-bit dithered 96KHz upsampling.

I'm using 0.74. Any ideas?

 

[lan]

Try 24bit padded to 32bit.

 

[Glassman]

lan is right, Envy24's native format is 32bit, so you have to choose 24bit fixed point padded to 32bit.. use triangular dithering..

 

[fewtch]

Quote:

Originally posted by pbirkett So instead I tried it with Direct Sound v2 output instead, and it didnt crash, and it did indeed sound different. It sounded worse to me. There was no drop in quality, but it did impose a softness to the sound and lost a little bit of power and drive. I tried it for a few days before switching back to 16-bit dithered KS.

I find your description here interesting. You say that it 'imposed a softness to the sound' -- as if the sound should be hard or harsh normally? Could it be that you've gotten used to a "digital-y" sound (imposed hardness or HF hash) and actually dislike something lacking in 'digititis' because it's unfamiliar?

I get that same "softening" and increased sense of air, and am convinced that's the way the music really sounds (vs. the usual digital-ish hard edge).

 

[Iron_Dreamer]

Quote:

Originally posted by temhem 1. at 48 sounds open up, and sound stage widens 2. Again at 48, the higher voices become richer and more liquid. And that digital (metallic) sound most complain about becomes less pronounced.

I noticed similar effects from going to 46 from 44, note that this is only through my DI/O. I will use 48 from here on out. Are you guys using the "slow mode"? I'm not, since it makes the sound stutter under high CPU load.

 

[fewtch]

Quote:

Originally posted by Iron_Dreamer I noticed similar effects from going to 46 from 44, note that this is only through my DI/O. I will use 48 from here on out. Are you guys using the "slow mode"? I'm not, since it makes the sound stutter under high CPU load.

I am... high quality resampling is more important to me than a light CPU load, and I'll stop the music if I need to do something demanding of the CPU. Why not use the best possible resampling if you're gonna do it (is my philosophy anyway)...

Edit -- I've been trying to decide whether I like 88.2/24 or 96/24 better, because 88.2 uses half the CPU time that 96KHz does in "slow mode" (20% vs. 40%)!! Differences are very subtle, if any... at the moment I'm sticking with 96/24, if for no other reason than to just provide 96KHz to a 96KHz DAC (no thoughts about "could it sound better?").

 

[pbirkett]

Quote:

Originally posted by fewtch I find your description here interesting. You say that it 'imposed a softness to the sound' -- as if the sound should be hard or harsh normally? Could it be that you've gotten used to a "digital-y" sound (imposed hardness or HF hash) and actually dislike something lacking in 'digititis'? I get that same "softening" and increased sense of air, and would swear on a stack of bibles that's the way the music really sounds (vs. the usual digital-ish hard edge).

You could be right mate. I can see what you mean about it being a little less digital, and I have to admit to liking my music a bit rough and ready, especially when called for - goes much better with dance and rock music IMO as I think in my mind these genres are supposed to have a dirtier and grittier sound rather than too smooth. Thats not to say I find 16-bit dithered KS harsh and gritty though, just that I dont particularly get on with the softening the upsampling seems to bring.

Nevertheless, I'll try 24-bit KS padded to 32 with upsampling tomorrow (when I go out this morning, I aint gonna get back until tomorrow now), and see if it can give something of a decent compromise - I think I'd be happier to have the same drive in the bass regions and a slight refining of the sound as long as it didnt take away too much in the involvement stakes.

 

[fewtch]

Quote:

Originally posted by Glassman yes this is it you should use ASIO output plugin and 24bit dithered output for whatever you play.. why using 16bit output to 24bit card? we can do better so why not..

Just a note for those not on XP or Win2K... generally, the right thing to use for Win95, Win98, 98SE and ME is "WaveOut" -- skip the "Microsoft Sound Mapper" and select the correct device for sound output (should be in a drop-down list in the plugin options).

P.S. don't use DirectX output with Win9x either, it munges the sound. Recommended to always use a "WaveOut" plugin with the proper "device" setting selected to bypass Windows and send the sound directly to the card. ASIO would probably work too (with the right plugin) but it's not necessary.

 

[lan]

I don't believe there's a real answer to foobar upsampling. It all depends on your hardware and equipment. So just try it.

Before when I used to have Audiophile 24/96 soundcard, I enjoyed upsampling more. Now, I find that leaving it off sound better.