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 559168 times) previous topic - next topic
0 Members and 2 Guests are viewing this topic.

lossyWAV Development

Reply #1175
It was clear that the bitrate difference between -q 2 and -q 3 would increase, but it's more or less negligible.
lame3995o -Q1.7 --lowpass 17

 

lossyWAV Development

Reply #1176
It was clear that the bitrate difference between -q 2 and -q 3 would increase, but it's more or less negligible.
Looking at the rate of change of bitrate, I'm more inclined to leave -snr preset values as per v0.9.8c as it gives a more even spread of bitrate.
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 #1177
It was clear that the bitrate difference between -q 2 and -q 3 would increase, but it's more or less negligible.
Looking at the rate of change of bitrate, I'm more inclined to leave -snr preset values as per v0.9.8c as it gives a more even spread of bitrate.

Practically speaking I prefer the way it's done with 0.9.8c a little bit too.
lame3995o -Q1.7 --lowpass 17

lossyWAV Development

Reply #1178
It was clear that the bitrate difference between -q 2 and -q 3 would increase, but it's more or less negligible.
Looking at the rate of change of bitrate, I'm more inclined to leave -snr preset values as per v0.9.8c as it gives a more even spread of bitrate.
Practically speaking I prefer the way it's done with 0.9.8c a little bit too.
I've re-visited the -spf function for higher frequencies at longer FFT lengths.

was: spreading_function_string        : string[precalc_analyses*(spread_zones+2)-1]='22222-22223-22224-12235-12246-12357';
now:  spreading_function_string        : string[precalc_analyses*(spread_zones+2)-1]='22222-22223-22224-12234-12245-12356';

this gives the following for my 53 problem sample set:
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 v0.9.8c  |327kbps|364kbps|406kbps|445kbps|468kbps|500kbps|533kbps|564kbps|595kbps|624kbps|653kbps|
|---------------|-------|-------|-------|-------|-------|-------|-------|-------|-------|-------|-------|
| beta v0.9.8d2 |328kbps|365kbps|407kbps|446kbps|470kbps|501kbps|534kbps|565kbps|596kbps|626kbps|654kbps|
|---------------|-------|-------|-------|-------|-------|-------|-------|-------|-------|-------|-------|

10 Album Test Set:
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 v0.9.8d2 |298kbps|???kbps|???kbps|???kbps|???kbps|463kbps|???kbps|???kbps|???kbps|???kbps|639kbps|
|---------------|-------|-------|-------|-------|-------|-------|-------|-------|-------|-------|-------|


Attached is a detailed analysis of bitrate vs quality preset for each sample in my 53 problem sample set.
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 #1179
...
was: spreading_function_string        : string[precalc_analyses*(spread_zones+2)-1]='22222-22223-22224-12235-12246-12357';
now:  spreading_function_string        : string[precalc_analyses*(spread_zones+2)-1]='22222-22223-22224-12234-12245-12356';

this gives the following for my 53 problem sample set:

As always I'm more interested in the bitrate increase for regular music (or in the relation bitrate increase of regular vs. problem tracks with a warm welcome to settings where bitrate increase is higher for the problem tracks).
Anyway, in this case it looks like difference is next to nothing in either case. This slightly more defensive setting is more or less for free.
lame3995o -Q1.7 --lowpass 17

lossyWAV Development

Reply #1180
As always I'm more interested in the bitrate increase for regular music (or in the relation bitrate increase of regular vs. problem tracks with a warm welcome to settings where bitrate increase is higher for the problem tracks).
Anyway, in this case it looks like difference is next to nothing in either case. This slightly more defensive setting is more or less for free.
lossyWAV beta v0.9.8d 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 #1181
Thank you, Nick.
As you tidied up the code a bit: does that mean it is necessary to do another listening test? Or were it just very minor changes?

Other than that I don't know how the community feels about it, but I'd welcome to go final now.
It doesn't look that we'll have essential changes any more, and it also looks as if there isn't a real need for changes, covering now the bitrate range from ~270 kbps up to 500+ kbps in a useful way.
lame3995o -Q1.7 --lowpass 17

lossyWAV Development

Reply #1182
Thank you, Nick.
As you tidied up the code a bit: does that mean it is necessary to do another listening test? Or were it just very minor changes?

Other than that I don't know how the community feels about it, but I'd welcome to go final now.
It doesn't look that we'll have essential changes any more, and it also looks as if there isn't a real need for changes, covering now the bitrate range from ~270 kbps up to 500+ kbps in a useful way.
No further changes to the mechanics of the method, only slight speed-up changes.

My only area of concern is with the -help and -longhelp (more the -longhelp) with respect to level of detail required to allow the user to make informed decisions....

The changes to the -spf parameters will not necessitate any more listening tests, I think the circa 1kbps increase in bitrate across the range of quality presets can only improve quality.

If no dissenting voices are forthcoming with respect to quality issues or required modifications to the help pages, v1.0.0 will be released on 12th May 2008 at 20:31 (precisely..... ).
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 #1183
I can't wait to see that as certified news on the front page.  Awesome.

What about a logo?  If the icon I made is used in the v1.0 binaries, then the logo could probably be delayed indefinitely, at least until there is a website/sf.net page for lossyWAV.

lossyWAV Development

Reply #1184
Some players (fb2k, WV winamp plugin) can use correction files for WavPack to play .WV+.WVC files losslessly. It seems no player can use *.lwcdf.XYZ with *.lossy.XYZ (say, .lossy.flac and .lwcdf.flac). It is of no importance for me, but...

lossyWAV Development

Reply #1185
... v1.0.0 will be released on 12th May 2008 at 20:31 (precisely..... ).

Great.
I've hesitated for so long reencoding my entire collection, but I'll do it with v1.0.0.
Guess I'll say goodbye to lossless archiving, will use a unique final VHQ lossyFLAC version instead for all the tracks in my collection, and will never touch them again.
(Life will be easier then as I hate having to manually overwrite the automatically generated replay gain values for quite a series of tracks after reencoding).
lame3995o -Q1.7 --lowpass 17

lossyWAV Development

Reply #1186
I've hesitated for so long reencoding my entire collection, but I'll do it with v1.0.0.
Guess I'll say goodbye to lossless archiving, will use a unique final VHQ lossyFLAC version instead for all the tracks in my collection, and will never touch them again.
(Life will be easier then as I hate having to manually overwrite the automatically generated replay gain values for quite a series of tracks after reencoding).
I've probably transcoded that portion (about 60%) of my collection which I have ripped to FLAC on my server about 15 to 20 times as lossyWAV has been in development....

I'll still be keeping the FLAC though.... (!)
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 #1187
"My only area of concern is with the -help and -longhelp (more the -longhelp) with respect to level of detail required to allow the user to make informed decisions....

The changes to the -spf parameters will not necessitate any more listening tests, I think the circa 1kbps increase in bitrate across the range of quality presets can only improve quality.

If no dissenting voices are forthcoming with respect to quality issues or required modifications to the help pages, v1.0.0 will be released on 12th May 2008 at 20:31 (precisely..... )."
[/quote]

Great news and well done everybody who's been involved especially Nick.C & halb27 of course.

I think you're right that the help/longhelp need to be good and, well, helpful. Also the wiki needs to be up to date and accurate.

I'm sure that people will want to rip to lossy.wav or even straight to lossy.flac, lossy.wv or whatever, but I guess that's something for others to pick up on once it's gone live.

Once again, well done. It's a great achievement

lossyWAV Development

Reply #1188
...My only area of concern is with the -help and -longhelp (more the -longhelp) with respect to level of detail required to allow the user to make informed decisions....

I was thinking about this area.
As a result I suggest we don't use the kind of advanced options we had so far.
We have arrived at these very differentiating -q levels (already too many levels for some users), and with this I don't see any sense in using even the most basic advanced option -nts. Why not just use a corresponding -q level? With -snr it's the same thing. We have a pretty defensive -snr setting with each quality level, but on the other hand not a lot of bitrate can be saved when going less defensive. IMO it's pretty balanced. Keeping away -nts and -snr from the user has the advantage that there's no need for describing which I guess is a difficult job.
The only useful option for the user IMO is the -clips options though I agree it's pretty personal and not really important either. In case we address the clipping as a user option I suggest we just use something like -noclips which makes sure that no clipping occurs (as done with the high -q levels). This can easily be described. IMO it can be one of the standard options.
lame3995o -Q1.7 --lowpass 17

lossyWAV Development

Reply #1189
I was thinking about this area.
As a result I suggest we don't use the kind of advanced options we had so far.
We have arrived at these very differentiating -q levels (already too many levels for some users), and with this I don't see any sense in using even the most basic advanced option -nts. Why not just use a corresponding -q level? With -snr it's the same thing. We have a pretty defensive -snr setting with each quality level, but on the other hand not a lot of bitrate can be saved when going less defensive. IMO it's pretty balanced. Keeping away -nts and -snr from the user has the advantage that there's no need for describing which I guess is a difficult job.
The only useful option for the user IMO is the -clips options though I agree it's pretty personal and not really important either. In case we address the clipping as a user option I suggest we just use something like -noclips which makes sure that no clipping occurs (as done with the high -q levels). This can easily be described. IMO it can be one of the standard options.
I could always change the -q 0 to 10 to -q 0 to 1.0 with a default of 0.5.... Actually, the more I think about that, the more I like it - and it's different from lame, ogg vorbis, flac, etc.
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 #1190
I could always change the -q 0 to 10 to -q 0 to 1.0 with a default of 0.5.... Actually, the more I think about that, the more I like it - and it's different from lame, ogg vorbis, flac, etc.

I like this idea (and also halb27's re. keeping options simple)

By the way, looking forward to world lossyWAV day on 12th May. How does one celebrate this? By listening to 24 hours of music?

C.
PC = TAK + LossyWAV  ::  Portable = Opus (130)

lossyWAV Development

Reply #1191
I could always change the -q 0 to 10 to -q 0 to 1.0 with a default of 0.5.... Actually, the more I think about that, the more I like it - and it's different from lame, ogg vorbis, flac, etc.
I like this idea (and also halb27's re. keeping options simple)

By the way, looking forward to world lossyWAV day on 12th May. How does one celebrate this? By listening to 24 hours of music?

C.
Code: [Select]
Procedure Celebrate;
Begin
  Repeat
    Success:=Drink_Beer and not Spill_Beer;
  Until Success=False;
  Goto Bed;
End;
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 #1192
I could always change the -q 0 to 10 to -q 0 to 1.0 with a default of 0.5.... Actually, the more I think about that, the more I like it - and it's different from lame, ogg vorbis, flac, etc.

Hmmm, I'm afraid using something like -q 1.0 is emotionally assciated with 'full (100%) quality', -q 0.6 is pretty much below in quality (though in reality it's overkill quality), and -q 0.15 is pretty bad (though in reality it's excellent). The problem is the association with a percentage quality scale.
But maybe it's just me who looks at it this way. In a sense we have this problem also with the -q 0 ... 10 scale, but because of the lacking association with a percentage scale the problem is less severe IMO.
BTW your suggestion is similar to the Nero AAC quality scale - however I don't see a problem in having a similar scale as another encoder as long as the emotional quality associations are corresponding, at least in the mid quality scale range.
lame3995o -Q1.7 --lowpass 17

lossyWAV Development

Reply #1193
Hmmm, I'm afraid using something like -q 1.0 is emotionally assciated with 'full (100%) quality', -q 0.6 is pretty much below in quality (though in reality it's overkill quality), and -q 0.15 is pretty bad (though in reality it's excellent). The problem is the association with a percentage quality scale.
But maybe it's just me who looks at it this way. In a sense we have this problem also with the -q 0 ... 10 scale, but because of the lacking association with a percentage scale the problem is less severe IMO.
BTW your suggestion is similar to the Nero AAC quality scale - however I don't see a problem in having a similar scale as another encoder as long as the emotional quality associations are corresponding, at least in the mid quality scale range.
I take your point - 0.0 to 1.0 does indeed have an immediate association with 0% to 100%. I'll leave the -q scale as is.
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 #1194
I take your point - 0.0 to 1.0 does indeed have an immediate association with 0% to 100%. I'll leave the -q scale as is.

Please do  . Can I ask for a progress bar (or percentage) and an option to create a log file ? I can't catch/save that informative 'average'line. When processing my disc images it's nice to see the progress counting up and down and up again, but I haven't a clue where it's going to. Anyway, many thanks for the program.

lossyWAV Development

Reply #1195
I take your point - 0.0 to 1.0 does indeed have an immediate association with 0% to 100%. I'll leave the -q scale as is.
Please do  . Can I ask for a progress bar (or percentage) and an option to create a log file ? I can't catch/save that informative 'average'line. When processing my disc images it's nice to see the progress counting up and down and up again, but I haven't a clue where it's going to. Anyway, many thanks for the program.
A progress bar in what sense and to what purpose? I suppose I could append a line to a log file with a "-l <filepath\filename>" type parameter but what would you want it to contain? When running from the command line there is a progress output of sorts which doesn't include the %age complete, rather amount of data processed (duration of WAV file processed) and number of bits removed so far.

A bit more detail as to what you would want to see would be nice.
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 #1196
... Can I ask for a progress bar ...

Everybody has its own way of how to practically do the encoding. If you give foobar2000 a try as a GUI for encoding you get a nice progress bar (and a good GUI). You just have to configure lossyWAV once for foobar2000 usage which isn't hard if you have a batch file for doing the lossyFLAC stuff (or whatever way you use lossyWAV).
lame3995o -Q1.7 --lowpass 17

lossyWAV Development

Reply #1197
I was asked for where to download the samples I usually try as potential problem samples for lossyWAV.

I've uploaded them for a limited time on my webspace. They can be downloaded from here.

Most of the samples don't show up problems (to me) even at a low quality setting like -q 1.5.
Even those samples that aren't perfect have an excellent quality to me even at -q 1.5.
I can abx however few of the samples at a mid quality level like -q 4.
With the triangle sample I don't know really how to think about it. In a strict sense I can't abx it because my results are too poor. Sometimes I can't hear a problem at all (at -q 4). But there are times when I can repeatedly get results like 4/4 before I start producing more and more errors.

It would be nice to learn about other people's experience.
lame3995o -Q1.7 --lowpass 17

lossyWAV Development

Reply #1198
Everybody has its own way of how to practically do the encoding.

I agree with that, but a percentage figure counting up while processing (like flac does) is more informative than counting up and down and up again to 256 MB (..).
Foobar won't work on my old and slower pc. And is rather overkill just to use it for just the information
Well, forget about the progression bar; it's still a great program. And mareo and the batchfiles do their tasks finely.

lossyWAV Development

Reply #1199
A progress bar in what sense and to what purpose? I suppose I could append a line to a log file with a "-l <filepath\filename>" type parameter but what would you want it to contain? When running from the command line there is a progress output of sorts which doesn't include the %age complete, rather amount of data processed (duration of WAV file processed) and number of bits removed so far.
A bit more detail as to what you would want to see would be nice.

A progress %age would be more convenient to me while working with dos-screen-output and batchfiles than the up- and downcounting to just 256 MB. When processing a disc image of say 771 MB, it's useless to see the up and downcounting not knowing where it ends. The percentage like flac does is fine.
And while processing I see an 'average' line with info that I would like to save. I'm a list and reports man. Especially in the testperiods it's nice to see what lossywav did with my files. I know most people us xp and foobar, but I'm with win98 on a slower and older computer, which is sufficient for playing my music and surfing.