Wavosaur home Wavosaur
Wavosaur free audio editor support forum
 
 FAQFAQ   SearchSearch   MemberlistMemberlist   RegisterRegister 
 ProfileProfile   Log in to check your private messagesLog in to check your private messages   Log inLog in 

Downsampling and reducing depth

 
Post new topic   Reply to topic    Wavosaur Forum Index -> General
 
Author Message
bolewin



Joined: 13 Jan 2008
Posts: 1
Location: Sweden

PostPosted: Sun Jan 13, 2008 5:54 pm    Post subject: Downsampling and reducing depth Reply with quote

For starters: Thanks for a capable and robust program.

I am using a Zoom H2 for recording which is usually done at 48 kHz (since the circuitry in the H2 is supposedly optimised for that although recording is possible also on 41.1 and 96 kHz) using 24 bits since the increased resolution makes it less imperative to be that dangerously close to 0dB.

Now to my queries, could anyone be so kind to comment on whether it is advisable to FIRST downsample to 41.1 kHz and THEN reduce depth from 24 to 16 bits, or if the other way around would be preferable?

I would also very much appreciate comments on the alghoritms used in Wavosaur for these procedures. Under which circumstances would the arious alternatives be preferable?

Looking forward to comments from authors and community!
Back to top
View user's profile Send private message
Rex
Site Admin


Joined: 05 Oct 2006
Posts: 797

PostPosted: Sun Jan 13, 2008 11:11 pm    Post subject: Reply with quote

Hello, welcome to the forum.

I think you can resample and bit-depth convert no matter the order.

I have to ask the main dev about the algorithms, there's a checkbox in Wavosaur resample tool for "filter" when downsampling, but i don't know much more.

For the bit depth convert, the 24 bits -> 16 bits reduction is done without dithering (it's in the todo list to add dither options).
I think you can use dithering VST plugins if it's really needed.
Back to top
View user's profile Send private message
PeterB



Joined: 19 Jun 2008
Posts: 3
Location: UK

PostPosted: Thu Jun 19, 2008 8:29 am    Post subject: Reply with quote

Thank you for Wavosaur. It is truly wonderful, and does exactly what I have been seeking for years!
I would be grateful however for an “Idiots Guide” to the cleanest way to convert 24/96 wav files to 16/44.1.
I have tried several ways to do this, some more successful than others, but none with entirely satisfying results. I also have tried using the MDA Dither and some other dithering VSTs , but I’m not sure I’m using them correctly.
Many thanks for any guidance!
Back to top
View user's profile Send private message
Rex
Site Admin


Joined: 05 Oct 2006
Posts: 797

PostPosted: Thu Jun 19, 2008 5:59 pm    Post subject: Reply with quote

Hello peter!

You can use the "resample" feature in the Process menu, but i suggest you to wait for the next release (we have fixed a bug with the filter/interpolation that wasn't active even is you check the "filter/interpolation" box )

I'll make some test with MDA dither and post the results in this thread
Back to top
View user's profile Send private message
PeterB



Joined: 19 Jun 2008
Posts: 3
Location: UK

PostPosted: Fri Jun 20, 2008 9:24 am    Post subject: Reply with quote

Thanks!
Back to top
View user's profile Send private message
Rex
Site Admin


Joined: 05 Oct 2006
Posts: 797

PostPosted: Tue Jun 24, 2008 12:15 pm    Post subject: Reply with quote

The procedure may look like this:

- load MDA dither in the VST rack, choose the bit depth destination, and the correct setting depending on your files, here's the help for MDA dither : http://mda.smartelectronix.com/vst/help/dither.htm

- apply the dithering then use menu->process->bit depth convert to set the bit depth to the "destination" value. (check "process quantization")
Back to top
View user's profile Send private message
Wavosaur Main Developer
Site Admin


Joined: 30 Sep 2006
Posts: 505
Location: France

PostPosted: Tue Jun 24, 2008 7:20 pm    Post subject: Reply with quote

Our resample dialog box has a bug: unfortunately the checkbox value is not read, and then the filter/interpolation we coded are always ignored.

Next release will correct this bug, with added, different choices for interpolation and anti alias filtering. Our DSP (Digital signal processing) has worked on this.

For bit depth converter we are working on dithering (ut not for next release...sorry).

For your information for the musicians we know, it's prefered to resample first and after to apply bit convertion.

But it's not a standard and I think each guys has its own habits.
_________________
~~~~~~~~~~~~~~~~~~
Wavosaur Main Developer
~~~~~~~~~~~~~~~~~~
http://www.wavosaur.com
Back to top
View user's profile Send private message Send e-mail Visit poster's website
acousmad



Joined: 14 Apr 2007
Posts: 6
Location: France

PostPosted: Fri Jun 27, 2008 10:49 am    Post subject: Reply with quote

Quote:
Our resample dialog box has a bug: unfortunately the checkbox value is not read, and then the filter/interpolation we coded are always ignored.
Next release will correct this bug, with added, different choices for interpolation and anti alias filtering. Our DSP (Digital signal processing) has worked on this.

I liked to use Wavosaur in the lessons I give for an example of how horrible can be aliasing when changing frequency rate...
Look at this :
Original 48 kHz file :


Resampled at 44,1 kHz in Wavosaur :


Now, I will be able to show them the importance of a good filter directly in Wavosaur Wink
Nice !

Quote:
For your information for the musicians we know, it's prefered to resample first and after to apply bit convertion.

Yes, in theory it is better since there will be less quantization errors during the resampling process.
Back to top
View user's profile Send private message
Wavosaur Main Developer
Site Admin


Joined: 30 Sep 2006
Posts: 505
Location: France

PostPosted: Fri Jun 27, 2008 6:03 pm    Post subject: Reply with quote

Yes, good analysis acousmad.

To be honest we saw the resampling "checkbox bug" with this type of drawning (sonogram) you presented here.

In the next release you will find different option to customize your resampling. And we are already sarching other options, but for future!
_________________
~~~~~~~~~~~~~~~~~~
Wavosaur Main Developer
~~~~~~~~~~~~~~~~~~
http://www.wavosaur.com
Back to top
View user's profile Send private message Send e-mail Visit poster's website
acousmad



Joined: 14 Apr 2007
Posts: 6
Location: France

PostPosted: Sat Sep 20, 2008 1:12 pm    Post subject: Reply with quote

Hi,

Hum... how to tell it...

The resampling result is always very bad with 1.0.3, to say the least...
With no interpolation, it goes from bad to very bad, and according to interpolation methods, it can go up to astonishing bad !
BTW, with or without antialiasing gives the same result.

The good thing is that Wavosaur remains my best software to show how resampling can destroy the sound Wink but I suppose that there is some people who would be happy to do a clean resampling with it ?

Try with a sine wave sliding from 20 to 20kH and resample it from 44,1 to 48 or the other way, and listen and look at the sonogram.

Any idea ?
Back to top
View user's profile Send private message
Wavosaur Main Developer
Site Admin


Joined: 30 Sep 2006
Posts: 505
Location: France

PostPosted: Sat Sep 20, 2008 1:59 pm    Post subject: Reply with quote

First the title of this post is downsampling and reducing depth.
So, if you generate a sinus from 48kHz to 44,1kHz, or 96kHz to 44,1Hz, you will see the difference with 1.0.2.0. For better result try bandlimited.

For upsampling, for the moment you could try different type of interpolations, excepted bandlimited.

Then it's not bad: it's just the result of algo types.

It's possible that in the future we will work on different algos.
_________________
~~~~~~~~~~~~~~~~~~
Wavosaur Main Developer
~~~~~~~~~~~~~~~~~~
http://www.wavosaur.com
Back to top
View user's profile Send private message Send e-mail Visit poster's website
tommyD



Joined: 03 Mar 2008
Posts: 57
Location: on The One

PostPosted: Sun Sep 21, 2008 1:58 pm    Post subject: Reply with quote

Re: sample rate conversion:

Rex wrote:
You can use the "resample" feature in the Process menu, but i suggest you to wait for the next release

I look forward to the next release...but in the meantime I recommend Voxengo R8brain (available here: http://www.voxengo.com/product/r8brain/). It's free, it offers batch processing, and I've obtained good results with it.

Hope that helps,

td

EDIT: ...And according to The Portable Freeware Collection, r8brain is portable. So like Wavosaur you can run it from a USB memory stick (see http://www.portablefreeware.com/?id=196#comments).
Back to top
View user's profile Send private message
tommyD



Joined: 03 Mar 2008
Posts: 57
Location: on The One

PostPosted: Sun Sep 21, 2008 5:12 pm    Post subject: Reply with quote

tommyD wrote:
r8brain is portable

Erm...kind of. I may have spoken too soon (sorry!).

You can certainly run it from a USB memory stick, but see my comment at http://www.portablefreeware.com/?id=196#comments
Back to top
View user's profile Send private message
EvilDragon



Joined: 14 May 2013
Posts: 11

PostPosted: Thu Oct 08, 2015 2:16 pm    Post subject: Reply with quote

Even after all these years (it's 7 years since this thread), upsampling is still horrible in Wavosaur...


Why not implement SoX resampler so everything gets perfectly resampled without annoying artifacts?


Previewing 44.1 files when audio interface is at 48k sounds absolutely horrible. Yes, I'm on latest version of Wavosaur (1.1.0.0, which is two years old by now)...
Back to top
View user's profile Send private message
Display posts from previous:   
Post new topic   Reply to topic    Wavosaur Forum Index -> General All times are GMT
Page 1 of 1

 
Other topics :





Powered by phpBB © 2001, 2005 phpBB Group