Skip to main content ↑ ↓

Notice

Please note that most of the software linked on this forum is likely to be safe to use. If you are unsure, feel free to ask in the relevant topics, or send a private message to an administrator or moderator. To help curb the problems of false positives, or in the event that you do find actual malware, you can contribute through the article linked here.
Topic: lossyWAV Development (Read 570569 times) previous topic - next topic
0 Members and 1 Guest are viewing this topic.

lossyWAV Development

Reply #1150 –
Quote from: halb27 on
Hallo Nick,

Bitrate increase of problem samples is welcome.
I always wonder in the first place what's the bitrate increase of regular music.
My personal opinion is that we should be very defensive towards the HF region with the short FFTs only.

Thank you for v0.9.7b. I'm curious about the bitrate with regular music and the quality of this version.
We've got visitors until Monday, but I'll try to get some of my 10 album test set results out tonight / tomorrow.
lossyWAV -q X -a 4 -s h -A --feedback 2 --limit 15848 --scale 0.5 | FLAC -5 -e -p -b 512 -P=4096 -S- (having set foobar to output 24-bit PCM; scaling by 0.5 gives the ANS headroom to work)

lossyWAV Development

Reply #1151 –
Quote from: Nick.C on
We've got visitors until Monday, but I'll try to get some of my 10 album test set results out tonight / tomorrow.

Take your time, don't hurry.
lame3995o -Q1.7 --lowpass 17

lossyWAV Development

Reply #1152 –
I did some listening tests using v0.9.7b -impulse -spf 22222-22223-22224-12235-12246-12357.
At -q 0 I could abx the difference against v0.9.7 with eig, but I wouldn't call it a significant improvement.
At -q 1 I couldn't, quality is high in both cases.

Bitrate for my regular track set goes up from 427 kbps to 442 kbps at -q 5 which I agree is way too high and could only be justified by a remarkable quality improvement.

I tested those samples from my last listening test which weren't totally correct.
I used v0.9.7 -q 3 -impulse.
Everything was fine to me with the exception of triangle-2_1644ds, though my abxing wasn't very good (7/10). Comparing this with v0.9.7b -q 3 -impulse -spf 22222-22223-22224-12235-12246-12357 I found that the encoding is identical to that of v0.9.7 -q 3 -impulse.

Judging from the results available so far I do think -impulse has a favorable effect. But it's fine the way you do it with v0.9.7, and the 22222 approach isn't efficient just as you said (I was just curious).
Guess I'll do some more listening tests tomorrow in order to confirm the positive effect of -impulse (hoping that everything will be alright now at -q 5 or even -q 4).
lame3995o -Q1.7 --lowpass 17

lossyWAV Development

Reply #1153 –
Quote from: Nick.C on
The problem seems to be with either foobar2000 or cmd.exe (I never did determine which). As soon as I removed spaces from the path to the batch file everything began to work.

Unicode handling in which sense?


If there's spaces in a static path in a batch file, it's easy enough to deal with by wrapping double-quotes around the path\executable parth of the command.  Such as:
Code: [Select]
@echo off
"c:\OMG S P A C E S\bin\lossyWAV.exe" %1 %3 %4 %5 %6 %7 %8 %9 -low -nowarn -quiet
"c:\OMG S P A C E S\bin\flac.exe" -5 -f -b 512 "%~N1.lossy.wav" -o "%~N2.flac"
del "%~N1.lossy.wav"


But to handle that with a variable as a path, it's a bit more complicated but fine once you know how to do left() and right() type functions with variables.

As a nice side effect of "massaging" the variables that way, you can handle also any other chars in the path, as well as in other variables such as those you might use for tagging.  Unicode for the tagging is mainly what I would think it's most useful for.

lossyWAV Development

Reply #1154 –
Quote from: jesseg on
If there's spaces in a static path in a batch file, it's easy enough to deal with by wrapping double-quotes around the path\executable parth of the command.  Such as:
Code: [Select]
@echo off
"c:\OMG S P A C E S\bin\lossyWAV.exe" %1 %3 %4 %5 %6 %7 %8 %9 -low -nowarn -quiet
"c:\OMG S P A C E S\bin\flac.exe" -5 -f -b 512 "%~N1.lossy.wav" -o "%~N2.flac"
del "%~N1.lossy.wav"
But to handle that with a variable as a path, it's a bit more complicated but fine once you know how to do left() and right() type functions with variables.

As a nice side effect of "massaging" the variables that way, you can handle also any other chars in the path, as well as in other variables such as those you might use for tagging.  Unicode for the tagging is mainly what I would think it's most useful for.
I have previously used spaces in the path-to-exe-files in the batch file, surrounded by double quotes as you mentioned - I just don't from preference as I keep the batch file in the same directory as the two exe files, but if there are spaces in the path to the batch file itself there seems to be a problem between foobar2000 and cmd.exe.
lossyWAV -q X -a 4 -s h -A --feedback 2 --limit 15848 --scale 0.5 | FLAC -5 -e -p -b 512 -P=4096 -S- (having set foobar to output 24-bit PCM; scaling by 0.5 gives the ANS headroom to work)

lossyWAV Development

Reply #1155 –
You could have look at Speek's batchencoder at Speek's Frontends
It's one of his fabulous front ends, and it works great for me.

lossyWAV Development

Reply #1156 –
I've computed the bitrates [kbps] for my full length regular track set using various settings:

           v0.96         v0.97      v0.9.7 -impulse      v0.9.7b -impulse -spf 22222-....
-q 0        263            272                290                                297
-q 1        281            288                304                                313
-q 2        305            311                325                                334
-q 3        335            339                351                                363
-q 4        372            375                385                                399
-q 5        417            418                427                                442
-q 6        447            448                456                                471
-q 7        477            477                485                                500

I've done the same thing with my short length problem tracks set but it's not worth while giving it here: it's more or less the same relations, just at a higher bitrate level.
I hoped bitrate would increase more with the problems than with the regular music as I was used to this behavior at the time we had a stronger skewing at the low frequency edge.

I did a lot of listening tests in order to arrive at conclusions what IMO should be the way to go.
From the consequences (not the history of my tests):

a) At the low bitrate edge we shouldn't use the 32 sample FFTs IMO.
I listened to furious, eig, and castanets, comparing the results of v0.9.7b -q 0 -impulse -spf 22222-.... with those of v0.9.7 -q 1, as well as  v0.9.7b -q 1 -impulse -spf 22222-.... with those of v0.9.7 -q 2.
As the result IMO it's better to use the next quality level instead of using -impulse -spf 22222-..... with both requirung roughly the same bitrate. The quality result was pretty clear for furious, there was no real difference for castanets, and just eig -q 0 -impulse -spf 22222-..... was is a tiny bit better to me than -q 1, and when doing the same comparison with quality levels advanced by 1 the preference for the 32 FFT is gone for eig too.
So using -impulse -spf 22222-.... is advantegous in rare cases whereas using the next quality level without the 32 sample FFTs brings a more general improvement.
Mooreover you've arrived at very good quality, Nick, way below 300 kbps, and if we used the 32 sample FFTs we would be at ~300 kbps which I guess isn't too attractive for the low bitrate users.

b) This morning with fresh ears I listened to all my potentially problematic samples again using v0.9.7 -q 4 -impulse. Now I could hear problems again with badvilbel (9/10), castanets (7/10), triangle (7/10).
Trying v0.9.7b -q 4 -impulse -spf 22222-.... made badvilbel and castanets non-abxable to me. With triangle I found out none of the 2 variants used with the 32 samples FFT produced a different file than the one when the 32 samples FFT isn't used at all.
I did the same thing with -q 5 instead of -q 4. Now badvilbel and castanets are okay to me using just -impulse (no -spf 22222-...). With triangle it's the same thing however as with -q 4: 7/10.
So the triangle problem can't be tackled by the 32 sample FFT, some other problems can but it looks like it's necessary to use a 22222 spreading.

As for the new -snr values when looking at the bitrate table they seem not to have a vital effect from -q 5 upwards.
Can you please further increase the -snr values from -q 5 up (maybe starting with a more gentle increase already at -q 3 or -q 4), Nick? I wonder if the problems can be tackled by this. When looking at this as an alternative to the 32 sample FFT accepting say half of the bitrate increase of the 32 sample FFts there's room for  a very serious increase of the -snr value.
lame3995o -Q1.7 --lowpass 17

lossyWAV Development

Reply #1157 –
Quote from: collector on
You could have look at Speek's batchencoder at Speek's Frontends
It's one of his fabulous front ends, and it works great for me.

I too love his frontends particularly Multi frontend that I use all the time. I've been trying to get Batchenc to work with LossyWAV but I've obviously got something wrong. I wonder if you can help

I have batchenc and LossyWAV in the same directory and the output directory in Batchenc set to the same as the input directory but when I run it Batchenc places the output files in its own directory (IE the one that contains Batchenc and LossyWAV). The only way I can get it to put the output files where I want them is to use the -o parameter in LossyWAV. Can you help me with that?

Thanks

lossyWAV Development

Reply #1158 –
Quote from: halb27 on
I did a lot of listening tests in order to arrive at conclusions what IMO should be the way to go.
From the consequences (not the history of my tests):

a) At the low bitrate edge we shouldn't use the 32 sample FFTs IMO.
I listened to furious, eig, and castanets, comparing the results of v0.9.7b -q 0 -impulse -spf 22222-.... with those of v0.9.7 -q 1, as well as  v0.9.7b -q 1 -impulse -spf 22222-.... with those of v0.9.7 -q 2.
As the result IMO it's better to use the next quality level instead of using -impulse -spf 22222-..... with both requirung roughly the same bitrate. The quality result was pretty clear for furious, there was no real difference for castanets, and just eig -q 0 -impulse -spf 22222-..... was is a tiny bit better to me than -q 1, and when doing the same comparison with quality levels advanced by 1 the preference for the 32 FFT is gone for eig too.
So using -impulse -spf 22222-.... is advantegous in rare cases whereas using the next quality level without the 32 sample FFTs brings a more general improvement.
Mooreover you've arrived at very good quality, Nick, way below 300 kbps, and if we used the 32 sample FFTs we would be at ~300 kbps which I guess isn't too attractive for the low bitrate users.

b) This morning with fresh ears I listened to all my potentially problematic samples again using v0.9.7 -q 4 -impulse. Now I could hear problems again with badvilbel (9/10), castanets (7/10), triangle (7/10).
Trying v0.9.7b -q 4 -impulse -spf 22222-.... made badvilbel and castanets non-abxable to me. With triangle I found out none of the 2 variants used with the 32 samples FFT produced a different file than the one when the 32 samples FFT isn't used at all.
I did the same thing with -q 5 instead of -q 4. Now badvilbel and castanets are okay to me using just -impulse (no -spf 22222-...). With triangle it's the same thing however as with -q 4: 7/10.
So the triangle problem can't be tackled by the 32 sample FFT, some other problems can but it looks like it's necessary to use a 22222 spreading.

As for the new -snr values when looking at the bitrate table they seem not to have a vital effect from -q 5 upwards.
Can you please further increase the -snr values from -q 5 up (maybe starting with a more gentle increase already at -q 3 or -q 4), Nick? I wonder if the problems can be tackled by this. When looking at this as an alternative to the 32 sample FFT accepting say half of the bitrate increase of the 32 sample FFts there's room for  a very serious increase of the -snr value.
How do you envisage the -snr values changing? At v0.9.7 the quality related presets are:

Code: [Select]
  spreading_function_string         : string[precalc_analyses*(spread_zones+2)-1]='22222-22223-22224-12235-12246-12357';

  quality_noise_threshold_shifts    : array[0..Quality_Presets] of Double  = (20,16,12,8,4,0,-2.4,-4.8,-7.2,-9.6,-12);

v0.9.6:  quality_signal_to_noise_ratios    : array[0..Quality_Presets] of Double  = (16,17,18,19,20,21,22.8,24.6,26.4,28.2,30); //v0.9.6

v0.9.7:  quality_signal_to_noise_ratios    : array[0..Quality_Presets] of Double  = (18,18.87,19.81,20.8,21.86,23,24.21,25.51,26.91,28.4,30); //variant #1

  quality_clips_per_channel         : array[0..Quality_Presets] of Integer = (3,3,3,3,2,1,0,0,0,0,0);


Firstly, I propose that the 32 sample FFT spreading function will be 22222 rather than 22223 as if it is used at all, it should be at its most(?) effective.

Secondly, I propose that -snr be (18,19,20,21,22,25,28,31,34,37,40).

Bitrates for 53 problem sample:
Code: [Select]
|---------------|-------|-------|-------|-------|-------|-------|-------|-------|-------|-------|-------|
|   lossyWAV    | -q 0  | -q 1  | -q 2  | -q 3  | -q 4  | -q 5  | -q 6  | -q 7  | -q 8  | -q 9  | -q 10 |
|---------------|-------|-------|-------|-------|-------|-------|-------|-------|-------|-------|-------|
| beta v0.9.6   |318kbps|338kbps|364kbps|394kbps|431kbps|472kbps|500kbps|529kbps|557kbps|584kbps|611kbps|
|---------------|-------|-------|-------|-------|-------|-------|-------|-------|-------|-------|-------|
| beta v0.9.7   |327kbps|346kbps|370kbps|400kbps|435kbps|475kbps|502kbps|530kbps|557kbps|584kbps|611kbps|
|---------------|-------|-------|-------|-------|-------|-------|-------|-------|-------|-------|-------|
| beta v0.9.7 i |342kbps|360kbps|383kbps|412kbps|446kbps|485kbps|513kbps|540kbps|567kbps|594kbps|619kbps|
|---------------|-------|-------|-------|-------|-------|-------|-------|-------|-------|-------|-------|
| beta v0.9.8   |327kbps|347kbps|371kbps|400kbps|435kbps|479kbps|511kbps|543kbps|574kbps|604kbps|632kbps|
|---------------|-------|-------|-------|-------|-------|-------|-------|-------|-------|-------|-------|
| beta v0.9.8 i |346kbps|365kbps|390kbps|419kbps|454kbps|500kbps|533kbps|564kbps|595kbps|624kbps|653kbps|
|---------------|-------|-------|-------|-------|-------|-------|-------|-------|-------|-------|-------|
lossyWAV -q X -a 4 -s h -A --feedback 2 --limit 15848 --scale 0.5 | FLAC -5 -e -p -b 512 -P=4096 -S- (having set foobar to output 24-bit PCM; scaling by 0.5 gives the ANS headroom to work)

lossyWAV Development

Reply #1159 –
Quote from: Nick.C on
... Firstly, I propose that the 32 sample FFT spreading function will be 22222 rather than 22223 as if it is used at all, it should be at its most(?) effective. ...


This makes sense to me.

Quote from: Nick.C on
...Secondly, I propose that -snr be (18,19,20,21,22,25,28,31,34,37,40). ...


If this yields the bitrate of 0.9.8 in your table this is reasonable for me too. The details depend on what we will do with the 32 samples FFT, cause I think we can't add several defensive methods or make the existing ones more defensive without sacrificing too much bitrate while lettng -q 5 be the setting with -nts 0.

At the moment I'd like to play around a bit trying to bring the triangle issue (though it's a subtle one) down with -q 4 or at least -q 5, and I'd like to try this first with a higher -snr value.
For this purpose it would be kind if you could also make the -snr option temporarily available for the user, Nick.
lame3995o -Q1.7 --lowpass 17

lossyWAV Development

Reply #1160 –
Quote from: halb27 on
If this yields the bitrate of 0.9.8 in your table this is reasonable for me too. The details depend on what we will do with the 32 samples FFT, cause I think we can't add several defensive methods or make the existing ones more defensive without sacrificing too much bitrate while lettng -q 5 be the setting with -nts 0.

At the moment I'd like to play around a bit trying to bring the triangle issue (though it's a subtle one) down with -q 4 or at least -q 5, and I'd like to try this first with a higher -snr value.
For this purpose it would be kind if you could also make the -snr option temporarily available for the user, Nick.
Yes, those quality preset parameters refer to v0.9.8 and correspond to the achieved bitrates.

lossyWAV beta v0.9.8 attached to post #1 in this thread. (-nts and -snr parameters temorarily re-enabled).
lossyWAV -q X -a 4 -s h -A --feedback 2 --limit 15848 --scale 0.5 | FLAC -5 -e -p -b 512 -P=4096 -S- (having set foobar to output 24-bit PCM; scaling by 0.5 gives the ANS headroom to work)

lossyWAV Development

Reply #1161 –
Thank you for v0.9.8, Nick.

The bitrate table for my full length regular track set up to -q 7:

-q 0   272 kbps
-q 1   289 kbps
-q 2   311 kbps
-q 3   340 kbps
-q 4   376 kbps
-q 5   422 kbps
-q 6   454 kbps
-q 7   486 kbps

So from -q 5 up the increased -snr do have a better effect compared to the values of 0.9.7.
Up to -q 4 however the effect is more or less negligible.

If we do want to have the 32 samples FFT when using the higher quality settings (and I think we should - hopefully we've got rid of the perception of a changed pitch by this, and maybe a 64 samples FFT was really too long as the short FFT at least with very high quality demands in mind). Even if a 32 sample FFT isn't necessary for very high quality it's one more weapon in the set of additional defensive tools).
The question is at what quality level to start with the 32 samples FFT. I think we definitely should have it with our default -nts 0 setting of -q 5. -q 5 -impulse takes an average bitrate of 446 kbps with my regular test set. So for the quality settings lower than -q 5 we have to redefine parameters a bit in order to get a roughly equally spaced quality and bitrate scale.

I've played around a lot with the parameters, and listened to especially the low bitrate settings.
What I've found, Nick, doesn't correspond totally with your idea of having all the quality parameter increase when going from one quality level to the next. Technically however I think there isn't any problem: The various quality parameters are defined for the integer quality levels. For a non-integer quality level you can take the quality parameters of the next lower integer quality level with the exception of -nts for which you can do a linear interpolation. As a rough description. For more details see below.

My suggestion for the quality parameters of integer quality levels:

-q 0: v0.9.8 -q 0 (average bitrate for my full length regular track set: 272 kbps).

For the next quality level IMO the best quality increase per kbps is by increasing -snr significantly. I've listened to -q 1 with various -snr setting, and going directly to -snr 22 brings an astonishing good quality to eig and furious which otherwise do suffer a bit from these low bitrate settings (though other than with these samples quality is remarkably high).

-q 1: v0.9.8 -q 1 -snr 22 (average bitrate for my full length regular track set: 307 kbps).

Once at -snr 22 increasing -snr further isn't so important. What should be more in focus is decreasing the -nts value significantly from the rather high value of -q 1's
-nts 16:

-q 2: v0.9.8 -q 2 -snr 22 -nts 9 (average bitrate for my full length regular track set: 337 kbps).

For -q 3 -nts should get lower one more time, but as we've arrived already at a modest -nts value I think we should use -impulse with -q 3 (if not we can lower -nts a bit more):

-q 3: v0.9.8 -q 3 -snr 22 -nts 6 -impulse (average bitrate for my full length regular track set: 382 kbps).

-q 4 should lower the -nts value one more time:

-q 4: v0.9.8 -q 4 -nts 3 -impulse (average bitrate for my full length regular track set: 409 kbps).

-q 5: v0.9.8 -q 5 -impulse (average bitrate for my full length regular track set: 446 kbps).

-q 6: v0.9.8 -q 6 -impulse (average bitrate for my full length regular track set: 479 kbps).

and so on.

Thus -nts goes smoothly from 9 (-q 2) to 6 (-q 3) to 3 (-q 4) to 0 (-q 5).

I've tried triangle with this setting of -q 4, and I can't abx it.
I've also looked at the effect -snr has for triangle. Unfortunately -snr doesn't have an influence on bits-to-remove (only with extreme values for -snr), just as is the case with -impulse. A lower -nts value is the only thing that helps.

This is my suggestion.
Compared to what we have right now IMO this means an improved quality for -q 1 and -q 2 while having bitrate still pretty low. As for -q 3+ we have an improved quality for the high end demand though at the cost of a higher bitrate. Moreover I think there is more 'meaning' in the quality steps up to -q 5.

A remark on the continuous quality scale as only the basic approach as I think of it is described above.
For the -q 0...1  range I think there should be linear interpolation (or any other continuous variation) with -snr as well as -nts.
With the -impulse parameter we can't have a smooth transition, and starting -impulse with -q 3.0 necessarily makes a larger jump in bitrate when going from -q 2 to -q 3 (more dramatically when going from -q 2.9 to -q 3.0).
Looking at the quality level analogy of Vorbis it's a bit similar to the situation where Vorbis starts using a different stereo handling at certain quality levels.
lame3995o -Q1.7 --lowpass 17

lossyWAV Development

Reply #1162 –
Quote from: jesseg on
i use a custom made "uber-brute-force" version which packs each half KB section with multiple types, and then going into previous sections, to figure out what lengths and what compression types will give maximum compression, and yes it's VERY slow, especially with executables over 1MB


Where can one go about finding this version?  Googling uber brute force upx leads to nothing but this thread.
Copy Restriction, Annulment, & Protection = C.R.A.P. -Supacon

lossyWAV Development

Reply #1163 –
Quote from: botface on
I've been trying to get Batchenc to work with LossyWAV but I've obviously got something wrong. I wonder if you can help
I have batchenc and LossyWAV in the same directory and the output directory in Batchenc set to the same as the input directory but when I run it Batchenc places the output files in its own directory (IE the one that contains Batchenc and LossyWAV). The only way I can get it to put the output files where I want them is to use the -o parameter in LossyWAV. Can you help me with that?

I had that too. But if it isn't a problem to use the -o parameter do so. The only thing is you have to change (fill in) the output directory name in the command line.
My command line for testing with lossyWav is "lossywav.exe <infile> -q 3 -force -o E:\iPOD". I admit it's more convenient when the programm uses <<output directory same as input directory >>. I'll experiment with that

[edit] A normal command line can be "FOR %%X IN ("*".WAV) DO LOSSYWAV "%%X" -q 3", without the outer quotes of course. [/edit]

lossyWAV Development

Reply #1164 –
Quote from: collector on
Quote from: botface on

I've been trying to get Batchenc to work with LossyWAV but I've obviously got something wrong. I wonder if you can help
I have batchenc and LossyWAV in the same directory and the output directory in Batchenc set to the same as the input directory but when I run it Batchenc places the output files in its own directory (IE the one that contains Batchenc and LossyWAV). The only way I can get it to put the output files where I want them is to use the -o parameter in LossyWAV. Can you help me with that?

I had that too. But if it isn't a problem to use the -o parameter do so. The only thing is you have to change (fill in) the output directory name in the command line.
My command line for testing with lossyWav is "lossywav.exe <infile> -q 3 -force -o E:\iPOD". I admit it's more convenient when the programm uses <<output directory same as input directory >>. I'll experiment with that

[edit] A normal command line can be "FOR %%X IN ("*".WAV) DO LOSSYWAV "%%X" -q 3", without the outer quotes of course. [/edit]

Thanks for responding. My command line is pretty much like yours. While it's not that much hassle having to specify the output directory I often find myself wanting to convert several albums at once, each one being in a separate input directory and wanting to keep them separate. I assumed I was doing something wrong because Multi Frontend, which is from the same developer and which I also use, will quite happily deal with a bunch of files from several different directories and put them back in the directories they originated from. Shame it can't handle LossyWAV .

lossyWAV Development

Reply #1165 –
Quote from: botface on
Thanks for responding. My command line is pretty much like yours. While it's not that much hassle having to specify the output directory I often find myself wanting to convert several albums at once, each one being in a separate input directory and wanting to keep them separate. I assumed I was doing something wrong
Shame it can't handle LossyWAV .

I am fond of those gadgets, batchfiles and front ends too. This afternoon I wrote myself a little batchfile that preserves the long filenames. For recursive actions I use the program 'sweep' together with it. It works and you can PM me if you're interested.

lossyWAV Development

Reply #1166 –
Quote from: halb27 on
My suggestion for the quality parameters of integer quality levels:

-q 0: v0.9.8 -q 0 (average bitrate for my full length regular track set: 272 kbps).

-q 1: v0.9.8 -q 1 -snr 22 (average bitrate for my full length regular track set: 307 kbps).

-q 2: v0.9.8 -q 2 -snr 22 -nts 9 (average bitrate for my full length regular track set: 337 kbps).

-q 3: v0.9.8 -q 3 -snr 22 -nts 6 -impulse (average bitrate for my full length regular track set: 382 kbps).

-q 4: v0.9.8 -q 4 -nts 3 -impulse (average bitrate for my full length regular track set: 409 kbps).

-q 5: v0.9.8 -q 5 -impulse (average bitrate for my full length regular track set: 446 kbps).

-q 6: v0.9.8 -q 6 -impulse (average bitrate for my full length regular track set: 479 kbps).

...I've tried triangle with this setting of -q 4, and I can't abx it....
I've implemented these changes into beta v0.9.8b and the bitrates for my 53 problem sample set are as follows:
Code: [Select]
|---------------|-------|-------|-------|-------|-------|-------|-------|-------|-------|-------|-------|
|   lossyWAV    | -q 0  | -q 1  | -q 2  | -q 3  | -q 4  | -q 5  | -q 6  | -q 7  | -q 8  | -q 9  | -q 10 |
|---------------|-------|-------|-------|-------|-------|-------|-------|-------|-------|-------|-------|
| beta v0.9.6   |318kbps|338kbps|364kbps|394kbps|431kbps|472kbps|500kbps|529kbps|557kbps|584kbps|611kbps|
|---------------|-------|-------|-------|-------|-------|-------|-------|-------|-------|-------|-------|
| beta v0.9.7   |327kbps|346kbps|370kbps|400kbps|435kbps|475kbps|502kbps|530kbps|557kbps|584kbps|611kbps|
|---------------|-------|-------|-------|-------|-------|-------|-------|-------|-------|-------|-------|
| beta v0.9.8   |327kbps|347kbps|371kbps|400kbps|435kbps|479kbps|511kbps|543kbps|574kbps|604kbps|632kbps|
|---------------|-------|-------|-------|-------|-------|-------|-------|-------|-------|-------|-------|
| beta v0.9.8b  |327kbps|364kbps|398kbps|438kbps|463kbps|500kbps|533kbps|564kbps|595kbps|624kbps|653kbps|
|---------------|-------|-------|-------|-------|-------|-------|-------|-------|-------|-------|-------|
Looking at the rate of change of bitrate, I might be tempted to increase the -nts on -3 slightly to "smooth out" a peak in the rate of change of bitrate, however these bitrates look fairly interesting....

lossyWAV beta v0.9.8b attached to post #1 in this thread.
lossyWAV -q X -a 4 -s h -A --feedback 2 --limit 15848 --scale 0.5 | FLAC -5 -e -p -b 512 -P=4096 -S- (having set foobar to output 24-bit PCM; scaling by 0.5 gives the ANS headroom to work)

lossyWAV Development

Reply #1167 –
Thank you, Nick.

It would be very nice if we could get some more listening feedback no matter at which quality scale.
lame3995o -Q1.7 --lowpass 17

lossyWAV Development

Reply #1169 –
It's not important, but I suggest a minor defensive change of -snr 22 to -snr 23.5 for -q 2 to -q 4.

My first motivation was a mere optical one:
With this the average bitrate for my regular tracks test set is 272 - 307 - 344 - 388 - 413 - 446 for -q 0 ... -q 5, which IMO is a tiny bit more equally spaced especially when going from -q 2 to -q 3 than 272 - 307 - 337 - 382 - 409 - 446.
Moreover we have a smooth -snr transition 22 - 23.5 - 25.

But I also think it's a good thing to have especially -q 2 more defensive while hardly sacrificing bitrate.
Sure between -q 1 and -q 2 -snr should vary as well as -nts. I think this makes an in-between quality level like -q 1.5 more attractive too.
lame3995o -Q1.7 --lowpass 17

lossyWAV Development

Reply #1170 –
Quote from: halb27 on
It's not important, but I suggest a minor defensive change of -snr 22 to -snr 23.5 for -q 2 to -q 4.

My first motivation was a mere optical one:
With this the average bitrate for my regular tracks test set is 272 - 307 - 344 - 388 - 413 - 446 for -q 0 ... -q 5, which IMO is a tiny bit more equally spaced especially when going from -q 2 to -q 3 than 272 - 307 - 337 - 382 - 409 - 446.
Moreover we have a smooth -snr transition 22 - 23.5 - 25.

But I also think it's a good thing to have especially -q 2 more defensive while hardly sacrificing bitrate.
Sure between -q 1 and -q 2 -snr should vary as well as -nts. I think this makes an in-between quality level like -q 1.5 more attractive too.
lossyWAV beta v0.9.8c attached to post #1 in this thread.
lossyWAV -q X -a 4 -s h -A --feedback 2 --limit 15848 --scale 0.5 | FLAC -5 -e -p -b 512 -P=4096 -S- (having set foobar to output 24-bit PCM; scaling by 0.5 gives the ANS headroom to work)

lossyWAV Development

Reply #1171 –
Thank you, Nick.
lame3995o -Q1.7 --lowpass 17

lossyWAV Development

Reply #1172 –
Quote from: halb27 on
Thank you, Nick.
I'm half tempted to change (18,22,23.5,23.5,23.5,25,28,31,34,37,40) to (18,22,22.75,23.5,24.25,25,28,31,34,37,40) as it appeals more to my linear tendencies....
lossyWAV -q X -a 4 -s h -A --feedback 2 --limit 15848 --scale 0.5 | FLAC -5 -e -p -b 512 -P=4096 -S- (having set foobar to output 24-bit PCM; scaling by 0.5 gives the ANS headroom to work)

lossyWAV Development

Reply #1174 –
Quote from: halb27 on
Quote from: Nick.C on
Quote from: halb27 on
Thank you, Nick.
I'm half tempted to change (18,22,23.5,23.5,23.5,25,28,31,34,37,40) to (18,22,22.75,23.5,24.25,25,28,31,34,37,40) as it appeals more to my linear tendencies....
This makes sense to me too.
Comparing the two:
Code: [Select]
|---------------|-------|-------|-------|-------|-------|-------|-------|-------|-------|-------|-------|
|   lossyWAV    | -q 0  | -q 1  | -q 2  | -q 3  | -q 4  | -q 5  | -q 6  | -q 7  | -q 8  | -q 9  | -q 10 |
|---------------|-------|-------|-------|-------|-------|-------|-------|-------|-------|-------|-------|
| beta v0.9.8b  |327kbps|364kbps|398kbps|438kbps|463kbps|500kbps|533kbps|564kbps|595kbps|624kbps|653kbps|
|---------------|-------|-------|-------|-------|-------|-------|-------|-------|-------|-------|-------|
| beta v0.9.8c  |327kbps|364kbps|406kbps|445kbps|468kbps|500kbps|533kbps|564kbps|595kbps|624kbps|653kbps|
|---------------|-------|-------|-------|-------|-------|-------|-------|-------|-------|-------|-------|
| beta v0.9.8d  |327kbps|364kbps|402kbps|445kbps|471kbps|500kbps|533kbps|564kbps|595kbps|624kbps|653kbps|
|---------------|-------|-------|-------|-------|-------|-------|-------|-------|-------|-------|-------|
lossyWAV -q X -a 4 -s h -A --feedback 2 --limit 15848 --scale 0.5 | FLAC -5 -e -p -b 512 -P=4096 -S- (having set foobar to output 24-bit PCM; scaling by 0.5 gives the ANS headroom to work)