Skip to main content
Topic: Yalac - Need some testing (Read 65246 times) previous topic - next topic
0 Members and 1 Guest are viewing this topic.

Yalac - Need some testing

Reply #75
Errors fixed.

My test corpus gives no errors anymore. I did check any preset with Yalacs internal verify option and performed an external binary compare of the decompressed files of the HIGH preset, which seemed to be the most affected.

Had to fix 3 errors. I suppose, that there still can be trouble with the last frame, if it is very small.

I'm not sure, if i should send the fixed version to the testers now, because i want do some more changes.

Only differences to the previous release:

- Errors fixed.
- Support for mono files.
- Some slight modifications for 24 bit files.
- Option to disable channel decorrelation (Joint stereo).

  Thomas

Yalac - Need some testing

Reply #76
Can we have a few pointers on what you're looking to change for the next release? (I'll soon be able to test it, btw..)

Yalac - Need some testing

Reply #77
Quote
' date='Apr 14 2006, 05:36 AM' post='382229']
Can we have a few pointers on what you're looking to change for the next release? (I'll soon be able to test it, btw..)


I hope to improve the speed of the fast and fastest presets a bit. It's more complex than i thought, so it can take some time.

Possibly it would be better to check for remaining errors before i think about further improvements (which will not be done within a couple of days).

What do you think?

Yalac - Need some testing

Reply #78
I hope to improve the speed of the fast and fastest presets a bit. It's more complex than i thought, so it can take some time.

Possibly it would be better to check for remaining errors before i think about further improvements (which will not be done within a couple of days).

What do you think?

In that case, maybe it's best to send everyone the newer version with presumably correct behavior.  That way, people will be able to make out errors if there are any.
Do you know if the verify function still sports false negatives?

If you need help thinking up optimizations, we're all here for you ;-)

You can get very technical (and even post some bits in german, this forum is very pluri-ethnic/multi-cultural) -- we can take the load.

Peace,
Tristan.

Yalac - Need some testing

Reply #79
Quote
' date='Apr 14 2006, 05:58 AM' post='382236']

I hope to improve the speed of the fast and fastest presets a bit. It's more complex than i thought, so it can take some time.

Possibly it would be better to check for remaining errors before i think about further improvements (which will not be done within a couple of days).

What do you think?

In that case, maybe it's best to send everyone the newer version with presumably correct behavior.  That way, people will be able to make out errors if there are any.


Agree. I hope to do it tomorrow äh... it's allready tomorrow (the birds just started singing). Should be done within the next 24 hours.

Quote
Do you know if the verify function still sports false negatives?


Not on my test corpus. But V0.03 definitely did.

Quote
If you need help thinking up optimizations, we're all here for you ;-)

You can get very technical (and even post some bits in german, this forum is very pluri-ethnic/multi-cultural) -- we can take the load.


Wow, got a hole new family here!

May need a new home soon. In 10 days they will dig for some world war two bombs in my neighbourship. I will have to leave home. Hope it will still be there, when i return...

  Thomas

Yalac - Need some testing

Reply #80
Agreed. I hope to do it tomorrow äh... it's allready tomorrow (the birds just started singing). Should be done within the next 24 hours.
Thomas


I must say, it is a great honour for me to know someone as dedicated to your software as yourself, and as energetic!

I truely wish you well in all your endeavours.  "May the hair never fall off your toes!"

Sleep well, 'gute nacht',
Tristan.

Yalac - Need some testing

Reply #81
Quote
' date='Apr 14 2006, 06:33 AM' post='382243']

Agreed. I hope to do it tomorrow äh... it's allready tomorrow (the birds just started singing). Should be done within the next 24 hours.
Thomas


I must say, it is a great honour for me to know someone as dedicated to your software as yourself, and as energetic!

I truely wish you well in all your endeavours.  "May the hair never fall off your toes!"

Sleep well, 'gute nacht',
Tristan.


A bit ashamed...

Thanks!

Good time to log out. It can't get better...

  Thomas

Yalac - Need some testing

Reply #82
YALAC Evaluation 0.04

for the testers.

Please look into the new history within readme for changes.

Important: Please specify the Yalac version in your comparisons!

Compression with 0.01 and 0.02 should be identical.

V0.03 and V0.04 both contain some small optimizations, that should help with some problem files, if Apply Window is enabled. Affects presets HIGH and INSANE.

It's very unlikely, that there soon will be more improvements of the compression efficiency. I have some ideas (your work has provided me with a new perspective on some details!), but they have to be implemented and tested, which should take at least some weeks.

Maybe that there will be some speed optimizations for HIGH in the meantime. I'm not sure.

Possible interim releases will only contain bug fixes.

Again, many thanks for your support!

  Thomas

Yalac - Need some testing

Reply #83
Errors fixed.

My test corpus gives no errors anymore. I did check any preset with Yalacs internal verify option and performed an external binary compare of the decompressed files of the HIGH preset, which seemed to be the most affected.

Had to fix 3 errors. I suppose, that there still can be trouble with the last frame, if it is very small.
This morning I have encode my test corpus using all presets, from Fastest to Insane, using v0.04.

Not one error.  I used the MD5 hash check as detailed in a previous post, and all hashes match against the source wave hashes.

Good news.
I'm on a horse.

Yalac - Need some testing

Reply #84
This morning I have encode my test corpus using all presets, from Fastest to Insane, using v0.04.

Not one error.  I used the MD5 hash check as detailed in a previous post, and all hashes match against the source wave hashes.

Good news.


I wasn't sure... and was very curious...

Glad that you tested so soon!

  Thomas

Yalac - Need some testing

Reply #85
somebody please split this thread into a separate "yalac names" thread.


 
SimplePortal 1.0.0 RC1 © 2008-2019