lossyWAV 1.2.0 released, Added noise WAV bitdepth reduction method Options

[Nick.C]

lossyWAV 1.2.0 is released.

lossyWAV is a near lossless audio processor which dynamically reduces the bitdepth of the signal on a block-by-block basis. Bitdepth reduction adds white noise to the processed output.

Changes from 1.1.0c:

• Now compatible with FFTW (requires libfftw3-3.dll to exist on the path);
• Spreading function changed;
• Noise shaping now off by default - can be switched on to automatic or user specified value;
• Introduction of --altpreset parameter which modifies internal presets and lowers upper frequency limit used in determining noise floor to 15.16kHz;
• Major code cleanup.

Download [maintenance release 1.2.0b]:

• Executable  lossyWAV_1.2.0b.zip ( 76.17K ) Number of downloads: 483
  
• Source  lossyWAV_Source_20101122_1.2.0b.zip ( 209.64K ) Number of downloads: 208
  

This post has been edited by Nick.C: Dec 2 2010, 21:48

[halb27]

Thank you, shadowking, for ABXing.
Not very good news, but valuable experience of lossyWAV behavior.
Do you mind publishing the sample and tell us the bad spot of the sample, shadowking?
I'd like to learn about transcoding quality whether it depends on -V.
Can you ABX the mp3 encoding of say -P --altpreset when you choose a better mp3 quality like -V2?

This post has been edited by halb27: Dec 21 2009, 14:14

[shadowking]

Thank you, shadowking, for ABXing.
Not very good news, but valuable experience of lossyWAV behavior.
Do you mind publishing the sample and tell us the bad spot of the sample, shadowking?
I'd like to learn about transcoding quality whether it depends on -V.
Can you ABX the mp3 encoding of say -P --altpreset when you choose a better mp3 quality like -V2?

Better news: I did more testing yesterday and as per your suggestion, I found out that the main problem was V5 itself - the samples i tested were already ruined by mp3 artifacts. LW didn't add any artifact that wasn't already there though the amplitude of it can change slightly. This is normal.

I tested V4 this time with plain Q1.5, 2.5, 3.5 without NS and things were way better. Q2.5 and over were fine. Q1.5 was abxable in one segment (lame V4 in itself wasn't fully transparent) but q1.5 added a bit of intensity to it. On subsequent trials in was somewhat harder to abx that part. Overall 1.5 was also very good - i test it as a 'low anchor'. Also tested Nero AAC q0.5 with lw Q2.5 and everything seemed transparent.

I'll do more tests soon and report them here.

[halb27]

Better news: ... Q2.5 and over were fine. ...

Thank you, shadowking. Though the --portable quality level does not necessarily have the property that transcoding to mp3 is fine it would have been a little bit disappointing.

[2Bdecided]

Better news: ... Q2.5 and over were fine. ...

Thank you, shadowking. Though the --portable quality level does not necessarily have the property that transcoding to mp3 is fine it would have been a little bit disappointing.

I think Nick's performed a minor miracle!

(Pre-processing with my original implementation changed the already audible artefacts on some samples when encoded to mp3 from much higher "lossyWAV" bitrates than you're seeing now.)

Cheers,
David.