Sample for Hybrid lossy formats Options

[shadowking]

Extreme tonal one from my cd collection. The 1st few seconds esp 1.1-2.0 produce hiss /noise using Wv lossy 4.5 @ 250 k . 300k is much better but still easy to abx - only apprent when volume is boosted. 350k is close - only a slight change in the 'color' of the background noise of the original. 400 k I was loosing focus on and off, but when it came too me I could abx 7/8 then 8/8 - at times I couldn't understand what the difference was or if there was even one present. I failed abxing several times @ 450 k even with the fast mode -fx3

I didn't test lossywav yet..

Attached File(s)

 Sick_Beds.flac ( 1.41MB ) Number of downloads: 126

 

[halb27]

... I didn't test lossywav yet..

Thanks for your sample. It would be great if you could try lossyWAV as well. Can't do it myself at the moment cause I'm sick and my hearing is suffering.

[shadowking]

I'll have to use Java lossywav as post 1.0 crashes wine for some reason..

BTW I tried -hx4 mode which Bryant likes :

350 k : Difficult 7/8
400 k : Fail 4/8


The slow -hx4 certainly does have quality control properties on certain samples and CPUs are getting fast (only affects encoding time). Its a little extra insurance when going for the < 400k bitrates. Though, at very high bitrate around 450 k the *fast* mode -fx4 is just as good with faster decoding / encoding.

This post has been edited by shadowking: Feb 9 2009, 14:16

[Nick.C]

Processed -P, -S, -E and -I, with detailed log file.

[edit] Removed as these files were processed with a development version of lossyWAV which has now been superseded. [/edit]

This post has been edited by Nick.C: Feb 24 2009, 21:56

[shadowking]

Processed -P, -S, -E and -I, with detailed log file.

Thanks for the files Nick , I'll post results soon.

[shadowking]

Lossywav -P : tiny fluctuation in hiss ?? failed abx 6/8

-P seems transparent here. No complaints !

[halb27]

Lossywav -P : tiny fluctuation in hiss ?? failed abx 6/8

-P seems transparent here. No complaints !

This sounds good. Thanks for testing.

[[JAZ]]

I'll have to use Java lossywav as post 1.0 crashes wine for some reason..

Ugh! The output of jLossyWav is still not the same than the one from Lossywav 1.1.0. Also, remember that it only accepts the -q parameter.

[Nick.C]

[JAZ] : These were processed with 1.1.2e with a minor mod to the spreading function - I wouldn't expect them to match 1.1.0b.

Very glad to see you still have enthusiasm for jlossyWAV!

This post has been edited by Nick.C: Feb 9 2009, 20:35

[[JAZ]]

Hehe. I spoke about 1.1.0b The output is very similar, but in some places it seems jLossywav adapts slower to the change (that, or it is less sensitive)

I am waiting to see how 1.1.2 evolves to start porting the changes once these issues are resolved.
I believe I've finally nailed the last things I didn't understand about the process and spread functions so I will add more comments to the code and describe the algorithm a bit.