avatar

tbrowne

Tin Man

Posts: 42 Member Since: 16/09/2012

Lead

Apr 3 16 6:52 PM

Tags : : , ,

Hello all,

It's been a while (about two and a half years, I reckon, due to full-time work), but at the start of the new year I was getting into DSD... the catch is that all my current replay stuff is PCM, and I wasn't convinced the freely available converters (and indeed, the not so free ones) were doing the best job of conversion... so I spent some time looking into the formats and thought, well, it wouldn't be that hard to support some other input rates as well... so here is the end result.

New features
- Supports 48,88.2,96,176.4,192,352.8,384,DSD64,DSD128 inputs, all with at least two filters - brickwall and gentle
- Gain switch
- Secret as-yet undocumented switch

It hasn't been hammered to heck, so I will not be surprised by the odd bug or two.  If you find one, please get in touch with a short sample file and the settings to recreate the problem... no guarantee when/if I'll be able to fix it, but it will be good to know.

A word on the filters... the /f0 default brickwall ones are basically all the same - as per the original 96kHz only support, it is roughly a 100Hz transition with an ultra flat passband before that.  The gentle filters are quite different, in comparison - generally getting gentler as the input sampling rate increases, so relying on low ultrasonic content to keep IMD down... some of them are generated via. different windowing methods, and one in particular is that ultimate purist goal... a Gaussian filter, which is active for 352.8kHz gentle.  Do use the gentle filters with caution, but with the right material, they are likely to prove the best-sounding choice.  Do not be afraid of a small amount of IMD, it doesn't have to ruin the sound of a recording... let your ears be the judge.

The DSD support... it works, but the default sharp filters are *very* slow.  If you think you have a fast computer, you are in for a shock... the gentle DSD filters are in comparison quite speedy.  I have found the results from the DSD downconversions to be very good using both kinds of filter... again, it depends on the material what works best for you.  There is a special "insane" setting for DSD64, which has a 50Hz passband and a deeper stopband... I may remove this filter in future as you are likely to have to run it all week to convert an album.  Note that no level compensation is performed on the DSD conversion... theoretically +3dB should be added but I found this resulted in clipping on some Sony SACDs... they must be running the modulator extremely hard (another free DSD converter clipped this album considerably).  On other recordings a straight 200% was preferable, so I have added an /x2 switch to facilitate this - this only really matters for those who go straight to 16-bit using FinalCD.

Download here.
http://www.sonicillusions.co.uk/downloads.htm
Hope it works well for you!

- Tom.
Quote    Reply   
avatar

ktm

Tin Man

Posts: 14 Member Since:14/12/2012

#6 [url]

Apr 6 16 12:03 PM

Hi Tom,
what are the output options for FinalCD 0.21 ?
Is it the same options as the input formats?
I´m on Mac and didn´t have time to check myself with tomase´s patch.
Thanks.
Kassian

Quote    Reply   
avatar

tbrowne

Tin Man

Posts: 42 Member Since:16/09/2012

#7 [url]

Apr 7 16 11:19 AM

Hi Kassian,
In terms of output options they haven't really changed... output is 44100Hz (hence the name FinalCD) at either 16 bit or 32 bit integer output wordlength. There is a little "flexibility" in terms of output rate, but I won't document this just yet as it is not well tested. ;)
Regards,
Tom.

Quote    Reply   
avatar

mgargel

Tin Man

Posts: 10 Member Since:02/09/2013

#9 [url]

Apr 10 16 4:55 AM

Hello Tom! and all!
I need help...
FinalCD is not working here and I can't figure out why. All it does is create an empty file (0KB).
I am using a MiniMac w/ bootcamp and Win7 64bit
Have you experienced this before?
Thanks in advance!

Quote    Reply   
avatar

tbrowne

Tin Man

Posts: 42 Member Since:16/09/2012

#10 [url]

Apr 10 16 3:01 PM

Hi there,

I am going to need more information! What are the exact contents of the command line given and what is the output in the console window?
If that doesn't yield an answer I will need a short sample to reproduce the problem with, although I am away at the moment (enjoyed some opera tonight!) so it may take a while to get to.
If you post the sample publicly, someone else might be able to test reproduction on another configuration in the meantime.

Regards,
Tom.

Quote    Reply   
avatar

apostolossiopis

Tin Man

Posts: 26 Member Since:21/02/2011

#12 [url]

Apr 11 16 3:26 PM

Using the same batch files I posted...
I have tested converting 3296000 and 2448000 files to 1644100 and 3244100 with sharp and gentle filters and the output was fine 

...so I can confirm that the batch files will work on windows XP for the above conversions using the latest FinalCD version (0.21) 
(I will have a go on my windows 7 machine possibly tomorrow)

again many thanks to Tom, as this software has become one of the most valuable tools in my methodology

Apostolos

Quote    Reply   
avatar

mgargel

Tin Man

Posts: 10 Member Since:02/09/2013

#13 [url]

Apr 11 16 3:38 PM

Apostolos, I don't think the batch files are the problem. In fact I think they are great and I thank you for that too!
I wonder if the 64bit OS is the problem... or perhaps the source file...



 

Quote    Reply   
avatar

tbrowne

Tin Man

Posts: 42 Member Since:16/09/2012

#14 [url]

Apr 12 16 2:53 AM

There's something about the source WAV file that it doesn't like... as far as I know, FinalCD will parse any legitimate WAV file, but I won't be able to know without the file. Post a link to it here so someone else can try it.

Quote    Reply   
avatar

apostolossiopis

Tin Man

Posts: 26 Member Since:21/02/2011

#15 [url]

Apr 12 16 10:24 AM

mgargel wrote:
Apostolos, I don't think the batch files are the problem. In fact I think they are great and I thank you for that too!
I wonder if the 64bit OS is the problem... or perhaps the source file...




 

I had to give it a go at some point, so it was a good reason to get on with it...

anyway, I would be happy to try out converting your file, if you post a link
 

Quote    Reply   
avatar

mgargel

Tin Man

Posts: 10 Member Since:02/09/2013

#16 [url]

Apr 12 16 12:27 PM

cool

here is the link to download the files: https://we.tl/gtYxYZaig2

You will find two files. Both are 1kHz tones. Interestingly one of them works fine...
The other (just like any audio I create in my computer) is the one FinalCD cannot understand. 

many many thanks for trying it
m.
 

Quote    Reply   
avatar

tbrowne

Tin Man

Posts: 42 Member Since:16/09/2012

#17 [url]

Apr 14 16 12:08 PM

Hi Mauricio,

I identified an issue with the 24-bit WAV file which I've modified the parser in FinalCD to cope with... Pro Tools inserts some strange sections at the start, and also an abnormally sized "fmt" section.  This did work fine in FinalCD 0.17, but got broken as a result of adding the DSD support which is predominantly 64 bit in chunks.

Please get the newest version 0.22 here...
http://www.sonicillusions.co.uk/downloads.htm

I've tested with your file and it converted with no problems, though it is entirely possible it has affected some other weird variation of WAV.  BTW, your 16 bit test file with a normal WAV header was under-dithered... something to be wary of!

Cheers,
Tom.

Quote    Reply   
avatar

viitalahde

Gold Finger

Posts: 597 Member Since:04/02/2011

#19 [url]

May 4 16 11:41 AM

Hello Tom, thank you so much for your work! I'm installing the new version, the older versions have been my to-go SRC's for quite some time now. I swear some records with something unpleasant in the quality of the high frequencies sound better at 44,1kHz done with your steep filters.

I'm still not seeing a "donate" button or anything like it on your website - I would be more than happy to contribute something for a good product.

Quote    Reply   
avatar

tbrowne

Tin Man

Posts: 42 Member Since:16/09/2012

#20 [url]

May 7 16 5:03 PM

Hello Jaakko,

That's very kind, thank you.  As I've said in the past, I have been uncomfortable about accepting donations as it can imply development will happen, although I will always consider adding a feature in return for payment if it matches the time required.  I prefer to dip in and out as there are many other audio projects to work on when I have spare time, like Sunrise and the 1-bit portable I'm ruminating on.  I have also developed a ruinous habit of 45rpm reissues since I got a Lyra for the turntable... maybe I should be accepting those... :)

It has been a lot of fun downsampling DSDs and "high-res" recordings to lower rates to see how they sound.  While I've had some very good results, I have yet to have a result as remarkable as the Dire Straits self-titled SHM CD (UICY-75769) which I listened to for the first time last night through the Discrete DAC, though... that is a stunning piece of work - my hats off to whoever did that transfer... to me it is better than my UK vinyl.  I have noticed that quite a few recordings improve with downsampling... I don't mean in terms usual of hi-fi niceties such as tone and air, I'm talking in terms of rhythm, timing... it is as if things have snapped into place more.  To me this is something that digital often gets wrong (and something I find older architectures of conversion seem to do better).

I have not ruled out that this is simply the fact that the 44.1 multiple clock is simply lower jitter in the DAC (this isn't even the best clocked implementation I've done), but I have a suspicion that something about the downsampling procedure that is removing... or more likely, masking something in the original conversion that is not entirely pleasant to the human ear.  If that could be nailed down, it would be a very powerful mastering tool.  I have considered writing a jitter injector, certainly...

Cheers,
Tom.

Quote    Reply   
Add Reply

Quick Reply

bbcode help