Hi Kay,
thanks again for the quick answer. I used *official* so I guess there is
some testing with the snapshot due.
Regarding testing with sndfile-resample, I tried that some time ago.
I didn't get the clicks but I didn't go up to 24bit as well.
So I was testing 16/44 (original)
against 16/96 ( sndfile-resample )
against 24/96 ( ecasound resample )
against 24/96 ( ecasound resample-hq )
I didn't find any other way to get my files from 16/44 to 24/96.
If there is, please let me know, so I can verify whether another
way of upsampling gives the expected results and we can compare
them to ecasound-resample-hq.
My current audiophiler devel machine is dead in the water, so I
won't have the time to set up the snapshot
##########################################################################
On Tue, 25 Jan 2005, Kamil Wencel wrote:
> Here is some overview :
> Latest ecasound
What version exactly, latest official or latest snapshot? Please
use the latest snapshot for any libsamplerate related testing:
http://ecasound.seul.org/download/snapshots/ecasound-2.3.6-cvs-20041221.tar.
gz
This is the same link Aaron posted earlier.
> This is important. Since there have been some issues in the past regarding
> standard libsamplerate, secret rabbit code and its resample-hq option in
> ecasound are a must. What was discovered is summarized by the following :
I agree. I won't release 2.4.0 before this is solved.
> Example 2: George Michael - Best of - Fastlove
[...]
> The results are catastrophic. Clearly hearable clicks and ticks
> especially
> when there is much dynamic in the song.
You could resample the song with sndfile-resample tool from libsamplerate
package and compare the result to file converted with Ecasound. You should
get identical results.
> Could this be some effect because the original file was compressed too
much
> during mastering ? this is quite a common problem these days. Most studios
> don't care about people with real high-end equipment anymore. They just
> stuff the music so that it will play well on radio or low tech customer
> gear.
This is possible. You could also try adding a -ea:95 to the processing
chain to see if the result sounds better. If yes, there is a bug somewhere
related to peak value processing...
> Someone on the list ( I don't remember his name ) had some ideas which
> sounded rather profound. Please let us not give up on that issue again.
I did major changes to libsamplerate support in the 2.3.6 snapshot linked
above.
-- http://www.eca.cx Audio software for Linux! -- To unsubscribe send message 'unsubscribe' in the body of the message to <ecasound-list-request@email-addr-hidden>. -- To unsubscribe send message 'unsubscribe' in the body of the message to <ecasound-list-request@email-addr-hidden>.Received on Wed Jan 26 00:16:04 2005
This archive was generated by hypermail 2.1.8 : Wed Jan 26 2005 - 00:16:05 EET