Computer Audio Asylum

Music servers and other computer based digital audio technologies.

Return to Computer Audio Asylum

Message Sort: Post Order or Asylum Reverse Threaded

ssrcx comparison on foobar 1.0.x

68.7.197.10
Posted on July 3, 2010 at 22:59:47
Adriel
Audiophile

Posts: 1037
Location: San Diego
Joined: October 13, 2001
 The SSRC High Precision / SSRC X plugin is the more obscure version, but contains:

foo_dsp_ssrcx.dll (works on foobar 1.0.)
foo_dsp_ssrc_highprec.dll (works on 0.9x)

Below are my resampling experiences with Foobar 0.9x, then the switch to SSRC X with Foobar 1.0.3.

I have mostly 44.1khz music. A few sacd, a few hd tracks, a few records. But mostly 44.1khz on the computer. Mostly poorly recorded rock, pop, that I do everything I can to make listenable.

With oversampling DACs I prefer no upsampling, as it looses detail when upsampling again. With non-oversampling DACS I prefer to upsample to 96khz (or downsample 192khz to 96khz for compatability).

With the older foobar 0.9 I tried a couple: sox with its various settings and mods, secret rabbit code, ssrc, and pphs.

-Secret rabbit code did the most to colour the sound, although in the end I preferred the somewhat thick and processed coloration.

-SSRC was the most neutral and accurate. I could downsample a hirez file and upsample it again, and SSRC made the result sound most like the original hirez file.

-Sox was artificial and fatiguing, whether minimum phase or linear.

-Pphs was soft and weak.

I recently updated to Foobar 1.0.3 and secret rabbit code stopped working.

So I ended up with one I hadn't tried yet, SSRC High Precision that works on Foobar 0.9x and its associated SSRC X that works on Foobar 1.0x.

The best presets for SSRC X are the otachan low, normal, high, top, and ultra settings. As quality increases, I notice AA (dB) increases, FFTFIRLEN increases, DF Hz decreases, 1M fs decreases.

I kept most of the otachan ultra settings. But I set FFTFIRLEN to 2097152. That is 8x the ultra setting. I tried 16x and beyond until foobar crashed, but 8x sounded best. That is fortunate, because the play lag only increases from there due to the cpu load.

I go for settings that give me the impression of higher resolution than there is, while balancing a smooth and fat sound, without inducing artificial tizz. And something that does not induce fatigue, since my system is tuned to be fatigue free.

I hope this helps out those on non-oversampling DACs and Foobar 1.0x, and who mainly have modern pop music.

 

Hide full thread outline!
    ...
RE: ssrcx comparison on foobar 1.0.x, posted on July 6, 2010 at 06:21:04
The Sound Guy
Audiophile

Posts: 716
Joined: November 16, 2003
 Thanks for this alternate to Sox.
Is there any documentation on the settings? AA, DF, 1M, etc?
When you select multiple sample, it outputs at the lowest one selected! Why?

 

Where can I download foo_dsp_ssrcx.dll ..., posted on July 5, 2010 at 13:18:39
cktc
Audiophile

Posts: 287
Location: So Cal.
Joined: May 7, 2005
 I also want to try.

 

RE: Where can I download foo_dsp_ssrcx.dll ..., posted on July 5, 2010 at 15:36:09
Adriel
Audiophile

Posts: 1037
Location: San Diego
Joined: October 13, 2001
 Hard to find the link, I will have to upload for you:

ssrcx.zip

 

RE: Where can I download foo_dsp_ssrcx.dll ..., posted on July 6, 2010 at 01:02:44
cktc
Audiophile

Posts: 287
Location: So Cal.
Joined: May 7, 2005
 Thank you, I have tried it. It really is system dependent. I like SOX more when upsampling to 24/96 to Sonic Frontiers SFD-2 MKIII. But I like SSRC X more when upsampling to 24/192 to TC Konnekt 8.

 

RE: ssrcx comparison on foobar 1.0.x, posted on July 4, 2010 at 23:40:30
fmak
Audiophile

Posts: 13158
Location: Kent
Joined: June 1, 2002
 The result you get with online src is more system dependent than if you upsample in the 'best' software mode first. I'd found Secret Rabbit to sound best on the old Foobar but I have upsampled all my 44.1 wav files using Audition 3.0.1 with no pre-post filter. HDDs are cheap!

 

RE: ssrcx comparison on foobar 1.0.x, posted on July 4, 2010 at 12:12:14
Tony Lauck
Audiophile

Posts: 13629
Location: Vermont
Joined: November 12, 2007
 The problem with the subjective evaluation of the SRCs is that you are listening to the combination of two filters, the one used to make the recording (anti-alias) and the one used to playback (anti-imaging). The optimum playback filter will depend on the recording filter. You can see this from the math, or you can run experiments with SRCs that allow you to play with the settings. (Although much more subtle, it's like the situation with LP equalization in the days prior to RIAA.)

If you try to do null tests (e.g. upsample and then downsample and then take the difference signal) you will generally get poor results, even if the SRCs are themselves accurate. There are two main reasons for this: (1) there may be a slight gain difference that will preclude a perfect null or (2) there may a timing difference. In doing null tests you can adjust for both if you can adjust for constant gain errors and if the timing delays are a integral number of samples. Unfortunately, many SRCs (such as iZotope 64 bit) have sub sample delays that make null tests difficult.

Note that you "should" be able to upsample and then downsample and get the same result (possibly differing only in the least significant bit) were it not for these glitches, assuming that the converter were perfect. Going the other way it will be impossible to get the same result (downsample and then upsample) on most recordings even with "perfect" converters, but the results may be close enough that they are aurally indistinguishable, depending on the recording and the resolution of your system.


Tony Lauck

"Diversity is the law of nature; no two entities in this universe are uniform." - P.R. Sarkar

 

Spot on about SoX. N/T, posted on July 4, 2010 at 09:54:46
carcass93
Audiophile

Posts: 7181
Location: NJ
Joined: September 20, 2006
 N/T

 

Have to say though I am using the oversampling, posted on July 3, 2010 at 23:35:45
ArdRi
 Resolution Audio Opus 21 as DAC (up to 24bit96kHz) I was again forced today to admit Cplay even with ASIO4All,SoX *VHQ*=96,Phase=Intermediate,bandwidth=90 Auto Buffer yet not even with cMP by a substantial degree bettered Foobar 1.0 viz Kernel Streaming playing a Redbook Virgin CD rip of Véronique Gens singing Berlioz.

There was instantaneous and clear recognition of reproduction of lower frequency from the Double Bass which had been imperceptible viz Foobar. All instrumental Timbre came through with fuller presence,accuracy and faithfulness to originality of tone as well as superior separation. It was so evident that I started it over in cMp and ran through the whole CD/rip. I'd felt Foobar was sounding cleaner than cPlay for a while but this revelation couldn't be ignored and this through a "Crap" Creative Tech Audigy 24 bit sound card..

 

Page processed in 0.024 seconds.

[ Top Top of Page ] [ Contact Us ] [ Support/Wish List ] [ Copyright © 2024, Audio Asylum, all rights reserved ]
[ General ] [ Speakers ] [ Tubes ] [ Vinyl ] [ Digital ] [ Hi-Rez ] [ Video Asylum ] [ Cables ] [ Tweaks/DIY ] [ Music ] [ Films ]