The best software for downsampling 192kHz and 96kHz ?

[J.W.]

What's the best software will the high quality sample rate conversion?

Is Samplitude downsampling algorithm any good? I have Sound Forge
7.0a but I've read that its sample rate conversion is only average.

I know that there are dedicated hardware units that work well but I'd
like to avoid that and stay strictly in the software realm. (Using
software to downsample also lets me convert wav files faster than
realtime.)

[Edi Zubovic]

On Tue, 10 Aug 2004 03:10:55 +0000, J.W. > wrote:

>
>
>What's the best software will the high quality sample rate conversion?
>
>Is Samplitude downsampling algorithm any good? I have Sound Forge
>7.0a but I've read that its sample rate conversion is only average.
>
>I know that there are dedicated hardware units that work well but I'd
>like to avoid that and stay strictly in the software realm. (Using
>software to downsample also lets me convert wav files faster than
>realtime.)
>

-- I don't know whether this one was the best but it's free and I
think worth trying.

http://www.voxengo.com/r8brain/

Edi Zubovic, Crikvenica, Croatia

[Johann Burkard]

J.W. wrote:
> What's the best software will the high quality sample rate conversion?

Here's another one: <http://www.mega-nerd.com/SRC/>

Johann
--
Meiner Meinung nach ist ein Reply-To-Posting ohne Einleitungszeile wie
ein Ei ohne Gelb!
(Uwe Premer in >)

[Arny Krueger]

"J.W." > wrote in message


> What's the best software will the high quality sample rate conversion?

> Is Samplitude downsampling algorithm any good? I have Sound Forge
> 7.0a but I've read that its sample rate conversion is only average.

You can test downsampling software yourself, both technically, and
subjectively.

An appropriate subjective test is a double blind listening test. You simply
take a recording that you think is critical, downsample it, upsample it, and
compare it to itself. If you can hear a difference, then there is something
wrong. You can find out more about the software and procedures for doing
this at www.pcabx.com .

One somewhat controersial way to subjectively compare sample rate
conversion, is to do the test above with a low-level passage, and then
amplify both the processsed and original music by some relatively large
amount such as 40 dB, and do the comparison. This might be a tougher test,
but since it uses artificial means to increase the sensitivity of the ear,
its not as easy to justify as a straight comparison.

Things to look for in your subjective evaluation include timbre changes and
an increase in noise, trashy sounds, and grit.

BTW if you have any doubts about it, this is not a fool's journey.
Downsampling programs have been found that audibly corrupt music.

An appropriate technical test involves the use of critical audio technical
analytical software such as the freeware Audio RightMark program. It has a
file-oriented mode of operation where you create a test file with the
program that you can process as you desire. You can then use the program to
analyze the processed file. The program does a fairly complete analysis and
provides you with a report. It will even write the HTML code so you can
publish your reports on the web.

Things to look for in your technical tests include changes in frequency
response, noise and dynamic range.

Total out-of-pocket costs: $0.00 .

> I know that there are dedicated hardware units that work well but I'd
> like to avoid that and stay strictly in the software realm. (Using
> software to downsample also lets me convert wav files faster than
> realtime.)

I understand that it is generally believed that the basic technical task of
resampling done in software, is conceptually and practically easier to do
than real-time hardware resampling, and therefore more likely to go well.
This is not to say that hardware resampling must be defective or inferior to
software resampling.

IME Audition/CE (with the quality turned up all the way, natch) does a very
good job of resampling. Performing the technical test mentioned above. the
results were (0.53 dB TPDF dither, unshaped dither, pre/post filtering,
quality = 999 )

CE 2.1
24/192 -> 16/44 -> 24/192

FR +0 -0.2 dB 20-20 KHz
Noise -118 dB A-weighted
Dynamic range 100 dB A-weighted
THD 0.0004%
IM (18 & 20 KHz) 0.0044%
Crosstalk -118 dB

Total time for test: less than 5 minutes.

That's about as good as 16/44 gets!

[Kurt Albershardt]

Edi Zubovic wrote:
> On Tue, 10 Aug 2004 03:10:55 +0000, J.W. > wrote:
>
>
>> What's the best software will the high quality sample rate conversion?
>>
>> Is Samplitude downsampling algorithm any good?
>
>
> I don't know whether this one was the best but it's free and I
> think worth trying.
>
> http://www.voxengo.com/r8brain/

Samplitude's SRC is no slouch, but...

Several posters on the Samplitude board have said that r8brain (when run in 64-bit mode) has the best SRC they have heard anywhere.

[J.W.]

On Tue, 10 Aug 2004 12:09:46 -0400, "Arny Krueger" >
wrote:

>
>To review:
>
>CE 2.1
>24/192 -> 16/44 -> 24/192
>
>FR +0 -0.2 dB 20-20 KHz
>Noise -118 dB A-weighted
>Dynamic range 100 dB A-weighted
>THD 0.0004%
>IM (18 & 20 KHz) 0.0044%
>Crosstalk -118 dB
>
>Notes - very straight-forward to use
>
>The new contender:
>
>R8Brain 24/96 -> 16/44 -> 24/96
>
>FR +0 -0.0 dB 20-20 KHz
>Noise -101.8 dB A-weighted
>Dynamic range 100 dB A-weighted
>THD 0.0004%
>IM (18 & 20 KHz) 0.0066%
>Crosstalk -103 dB
>
>Notes - limited file type support
>
>Comments: basically, pick one! ;-)
>


You compared CE 2.1 at 192kHz and R8Brain at 96kHz... was this
intentional? I thought r8brain could handle 192kHz.

[Arny Krueger]

"J.W." > wrote in message

> On Tue, 10 Aug 2004 12:09:46 -0400, "Arny Krueger" >
> wrote:
>
>>
>> To review:
>>
>> CE 2.1
>> 24/192 -> 16/44 -> 24/192
>>
>> FR +0 -0.2 dB 20-20 KHz
>> Noise -118 dB A-weighted
>> Dynamic range 100 dB A-weighted
>> THD 0.0004%
>> IM (18 & 20 KHz) 0.0044%
>> Crosstalk -118 dB
>>
>> Notes - very straight-forward to use
>>
>> The new contender:
>>
>> R8Brain 24/96 -> 16/44 -> 24/96
>>
>> FR +0 -0.0 dB 20-20 KHz
>> Noise -101.8 dB A-weighted
>> Dynamic range 100 dB A-weighted
>> THD 0.0004%
>> IM (18 & 20 KHz) 0.0066%
>> Crosstalk -103 dB
>>
>> Notes - limited file type support
>>
>> Comments: basically, pick one! ;-)

> You compared CE 2.1 at 192kHz and R8Brain at 96kHz... was this
> intentional?

Yes, I was having sample format issues. I later resolved them:

> I thought r8brain could handle 192kHz.

It can!

R8Brain 24/192 -> 16/44 -> 24/192

FR +0 -0.01 dB 20-20 KHz
Noise -101.9 dB A-weighted
Dynamic range 100.2 dB A-weighted
THD 0.0006%
IM (18 & 20 KHz) 0.0066%
Crosstalk -101.9 dB

BTW CE has some FR issues when the spread of sample rates gets very large,
which are just hinted at by the 44<->192 test results. R8Brain may have them
too, but they are less of an issue. Not that CE's situation is any kind of
an meaningful issue, except to people who are looking for perfect frequency
response and really big multiples of sample rate (e.g. 20).

[Geoff Wood]

J.W. wrote:
> What's the best software will the high quality sample rate conversion?
>
> Is Samplitude downsampling algorithm any good? I have Sound Forge
> 7.0a but I've read that its sample rate conversion is only average.

You probably heard that from a religous adherent to a different editor.
IIRC there are settings for conversion quality. Try it on 'Best' and see if
you like it.

Also, isn't down-convert to 1/2 Fs a fairly trivial procedure ?

geoff

[J.W.]

On Fri, 13 Aug 2004 08:56:56 +1200, "Geoff Wood"
-nospam> wrote:

>J.W. wrote:
>> What's the best software will the high quality sample rate conversion?
>>
>> Is Samplitude downsampling algorithm any good? I have Sound Forge
>> 7.0a but I've read that its sample rate conversion is only average.
>
>You probably heard that from a religous adherent to a different editor.
>IIRC there are settings for conversion quality. Try it on 'Best' and see if
>you like it.
>
>Also, isn't down-convert to 1/2 Fs a fairly trivial procedure ?

I'd be downsampling all the way from 192kHz down to 44.1kHz. More
than 1/4th the samples and it's not an exact multiple of 192.

[Geoff Wood]

J.W. wrote:
> On Fri, 13 Aug 2004 08:56:56 +1200, "Geoff Wood"
> -nospam> wrote:
>
>> J.W. wrote:
>>> What's the best software will the high quality sample rate
>>> conversion?
>>>
>>> Is Samplitude downsampling algorithm any good? I have Sound Forge
>>> 7.0a but I've read that its sample rate conversion is only average.
>>
>> You probably heard that from a religous adherent to a different
>> editor. IIRC there are settings for conversion quality. Try it on
>> 'Best' and see if you like it.
>>
>> Also, isn't down-convert to 1/2 Fs a fairly trivial procedure ?
>
> I'd be downsampling all the way from 192kHz down to 44.1kHz. More
> than 1/4th the samples and it's not an exact multiple of 192.

Sorry , I mis-read the title as "196K TO 96K" !

geoff

[Geoff Wood]

J.W. wrote:
> On Fri, 13 Aug 2004 08:56:56 +1200, "Geoff Wood"
> -nospam> wrote:
>
>> J.W. wrote:
>>> What's the best software will the high quality sample rate
>>> conversion?
>>>
>>> Is Samplitude downsampling algorithm any good? I have Sound Forge
>>> 7.0a but I've read that its sample rate conversion is only average.
>>
>> You probably heard that from a religous adherent to a different
>> editor. IIRC there are settings for conversion quality. Try it on
>> 'Best' and see if you like it.
>>
>> Also, isn't down-convert to 1/2 Fs a fairly trivial procedure ?
>
> I'd be downsampling all the way from 192kHz down to 44.1kHz. More
> than 1/4th the samples and it's not an exact multiple of 192.

Sorry , I mis-read the title as "196K TO 96K" !

geoff

[Arny Krueger]

"Geoff Wood" -nospam> wrote in message

> J.W. wrote:
>> What's the best software will the high quality sample rate
>> conversion?
>>
>> Is Samplitude downsampling algorithm any good? I have Sound Forge
>> 7.0a but I've read that its sample rate conversion is only average.
>
> You probably heard that from a religous adherent to a different
> editor. IIRC there are settings for conversion quality. Try it on
> 'Best' and see if you like it.
>
> Also, isn't down-convert to 1/2 Fs a fairly trivial procedure ?

Not really. The first step should be to brick-wall the source at a frequency
closely realated to 1/4 Fs.

[Arny Krueger]

"Geoff Wood" -nospam> wrote in message

> J.W. wrote:
>> What's the best software will the high quality sample rate
>> conversion?
>>
>> Is Samplitude downsampling algorithm any good? I have Sound Forge
>> 7.0a but I've read that its sample rate conversion is only average.
>
> You probably heard that from a religous adherent to a different
> editor. IIRC there are settings for conversion quality. Try it on
> 'Best' and see if you like it.
>
> Also, isn't down-convert to 1/2 Fs a fairly trivial procedure ?

Not really. The first step should be to brick-wall the source at a frequency
closely realated to 1/4 Fs.