8th September 2012
|
#1 | | Lives for gear
Joined: Aug 2004 Location: Germany
Posts: 3,419
Thread Starter  | Transparent upsampling?
A long time ago after comparing lot's of programs and plugins I sampled on Voxengo R8brain's "linear" algorithm for downsampling audio files. I never considered upsampling much of a problem, but I recently discovered (and confirmed it testing with a 1 sample spike and subsequently with music material) that the "linear" algorithm significantly, audibly 'blunts' the transients compared to the "minimum" algo when upsampling.
There's this nice comparison online of all the SRC software with frequency and impulse responses etc., but only for downsampling. Even so, none of the algorithms is transparent in that you get a 1 sample only impulse out of the input (which in all fairness doesn't even happen converting via hardware set to different samplerates).
For practical purposes, is there a sample rate conversion software (preferably Windows based) that does not audibly degrade the signal at all sampling up and down again?
| 
|  | |
8th September 2012
|
#2 | | Motown legend
Joined: Jun 2002 Location: Songwriter Gulch, Nashville TN
Posts: 12,057
Verified Member  |
Weiss Saracon
| |
| | |
8th September 2012
|
#3 | | Gear maniac
Joined: Oct 2007
Posts: 231
|
iZotope RX2 Advanced includes a sample rate converter (Dubbed SRC). Check it out.  | |
| | |
8th September 2012
|
#4 | | Gear maniac
Joined: Mar 2012 Location: Frankfurt/Germany | Quote:
Originally Posted by living sounds  For practical purposes, is there a sample rate conversion software (preferably Windows based) that does not audibly degrade the signal at all sampling up and down again? | No. They are all lossy "by design".
But the quality difference between different products is shocking. I also recommend the izotope src.
| |
| | |
8th September 2012
|
#5 | | Lives for gear
Joined: Aug 2004 Location: Brooklyn, New York
Posts: 3,834
Verified Member |
Good source for test comparisons at SRC Comparisons
Best regards,
Steve Berson
| |
| | |
9th September 2012
|
#6 | | Lives for gear
Joined: Aug 2004 Location: Germany
Posts: 3,419
Thread Starter |
Thanks, I'll check out the Izotope product! The SRC comparison website was what I was reffering to.
| |
| | |
9th September 2012
|
#7 | | Lives for gear
Joined: Nov 2006 Location: Seattle
Posts: 1,923
Verified Member |
For upsampling, the best I've heard is the apodizing filter in Pyramix
| |
| | |
9th September 2012
|
#8 | | Lives for gear
Joined: Dec 2007
Posts: 875
Verified Member | Quote:
Originally Posted by living sounds There's this nice comparison online of all the SRC software with frequency and impulse responses etc., but only for downsampling. | For most software, you can expect similar results for upsampling and downsampling. Quote:
Originally Posted by living sounds Even so, none of the algorithms is transparent in that you get a 1 sample only impulse out of the input | The "transparent" output for a 1-sample input is not necessarily a 1-sample. Outputs from most SRCs resemble a sinc function and they should all sound transparent. The ringing that you are seeing should not be audible as it only happens near the Nyquist frequency. Quote:
Originally Posted by living sounds For practical purposes, is there a sample rate conversion software (preferably Windows based) that does not audibly degrade the signal at all sampling up and down again? | Even if such software exists, it's quite likely that the quality of the upsampled signal will be mediocre. A paradox, I know...
| |
| | |
9th September 2012
|
#9 | | Gear addict
Joined: Oct 2010 Location: Miami & Los Angeles
Posts: 445
|
I trust both the Weiss Saracon and iZotope fully.
I prefer the workflow and features of Weiss Saracon though. The iZotope can give option anxiety, and honestly everyone I know uses the Saracon.
Have you ever used Weiss gear? The amount of care in the design really inspires confidence. I don't think anyone's ever had a complaint. . .
But anyway, as I was told by someone very intelligent who has all my respect recently, if the difference is marginal, always go with the less expensive.
Hm.
__________________ The Album as an Art Form
Writings on album production, songwriting, engineering, arranging, and more.
| |
| | |
9th September 2012
|
#10 | | Lives for gear
Joined: Nov 2006 Location: Seattle
Posts: 1,923
Verified Member |
Just don't use Saracon for any DSD work. I'm no longer using mine. What's Best Forum | |
| | |
9th September 2012
|
#11 | | Gear maniac
Joined: Dec 2010
Posts: 196
|
Lavry Gold 3000S best choice.
| |
| | |
9th September 2012
|
#12 | | Gear nut
Joined: Jan 2011
Posts: 102
| Quote:
Originally Posted by monsieur x I trust both the Weiss Saracon and iZotope fully.
I prefer the workflow and features of Weiss Saracon though. The iZotope can give option anxiety, and honestly everyone I know uses the Saracon.
Hm. | Price point considered or not, I also trust Saracon.
| |
| | |
10th September 2012
|
#13 | | Lives for gear
Joined: Aug 2003 Location: Hollywood CA
Posts: 3,010
Verified Member | Quote:
Originally Posted by MainTime Lavry Gold 3000S best choice. | Does not compare well to today's best software or hardware solutions, imo.
DC
| |
| | |
10th September 2012
|
#14 | | Lives for gear
Joined: Aug 2003 Location: Hollywood CA
Posts: 3,010
Verified Member | Quote:
Originally Posted by DSD_Mastering | How much aliasing is in the one you use?
DC
| |
| | |
10th September 2012
|
#15 | | Lives for gear
Joined: Nov 2006 Location: Seattle
Posts: 1,923
Verified Member | Quote:
Originally Posted by dcollins How much aliasing is in the one you use?
DC | I haven't measured it. I'm using custom filters that were written for us to use for DSD work in MatLab.
| |
| | |
10th September 2012
|
#16 | | Lives for gear
Joined: Aug 2003 Location: Hollywood CA
Posts: 3,010
Verified Member | Quote:
Originally Posted by DSD_Mastering I haven't measured it. I'm using custom filters that were written for us to use for DSD work in MatLab. | It might be worth investigating, as it's likely to have some.
DC
| |
| | | |
New Reply 
Facebook
Twitter
LinkedIn
