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 1.1.0 Development Thread. (Read 184455 times) previous topic - next topic
0 Members and 1 Guest are viewing this topic.

lossyWAV 1.1.0 Development Thread.

Reply #100 –
so if we mix, first we have something like this :

-q 00.0 = -nasty
-q 02.5 = -portable
-q 05.0 = -archive or -default
-q 07.5 = -overkill
-q 10.0 = -paranoid

it's not my ideal, but as I am not alone to use the codec ... it's better than nothing & I can live with it
& now if we kill -nasty & -paranoid, but agree that -paranoid is a better word than -overkill, we have.

-q 02.5 = -portable
-q 05.0 = -archive or -default
-q 07.5 = -paranoid

I like it like this.

Edit:
As -portable & -archive describe the use rather than the quality (unlike -nasty, -overkill & -paranoid)
maybe it would be better to use 3 words that all describe the final use, so:

-q 02.5 = -portable
-q 05.0 = -archive or -default
-q 07.5 = -transcoding

is possible too & even more self-speaking for noobs.

lossyWAV 1.1.0 Development Thread.

Reply #101 –
Quote from: shadowking on
I had another idea: create a few 'presets' that are mapped to common Q settings. Maybe even hide the Q scale from the normal screen and document it with --longhelp or similar.

-normal [Default] = Q5
-medium = Q3
-Portable = Q2

- Extreme / Archiving / transcoding = Q6..10

That's how I do it for the moment. My batchfile exists of three settings to test. Q9 for near-lossless archives, Q6 for playable flac images and Q3 for DAP. I mentioned earlier that imho the eleven step scale is too wide. Let alone the numerous steps in between. 5.5678 or 0.1234    To hear the differences quickly I take these bigger steps.
There's also the various combo's in shaping, detail, lowpass. A half-deaf noob would be going mad.

lossyWAV 1.1.0 Development Thread.

Reply #102 –
Quote from: halb27 on
[...], so it's hard to find a simple name for -q 10 (maybe -paranoid hits it best - but it's a pretty negative word. However if most members would appreciate a -nasty mode among the standard options a -paranoid mode would be a good counterweight).
Another argument for having just three standard quality options.

For -paranoid (or -riskless) I use flac.    And yes, maybe shaping and such should be an embedded part of the presets. so an inexperienced user won't have to bother about these settings. Like -e -r -b 4096 in flac; just -best or -fast.

lossyWAV 1.1.0 Development Thread.

Reply #103 –
Combining some previous suggestions posted here, these are the presets I'd like to see:

Code: [Select]
--q 2.5 = -portable
--q 5.0 = -normal
--q 7.5 = -archive

Of course, the presets would be mutually-exclusive with the "--q" advanced setting.
lossyFLAC (lossyWAV -q 0; FLAC -b 512 -e)

lossyWAV 1.1.0 Development Thread.

Reply #104 –
I wonder if -archive is a good term afterall as it really depends on what people intend to do with their "archive", personnaly I would keep -archive out of the game & keep it for lossless. Some people use lossy for archive because it's transparent to them, some people use lossless for archive as it is perfection to them. There is no consensus on it. So, I slowly tend toward:

-q 2.5 = -portable
-q 5.0 = -default
-q 7.5 = -transcoding

you could as well use -q 5 or -q7.5 for archival purpose as there is no evidence that transcoding -q 7.5 to lossy for DAP would be more transparent than transcoding -q 5.0 to lossy for DAP even if -q 7.5 is less agressive on paper.

lossyWAV 1.1.0 Development Thread.

Reply #105 –
So the main problem is whether the naming should address the application (like 'portable', 'archive', 'transcoding') or the quality.

Targeting at the application has its problems as we see in the discussion. I for instance wouldn't like -q 5 to be associated with archiving (archiving in the sense of a substitute for a lossless archive). I also don't like a specific 'transcoding' quality as I'd expect even from -q 2.5 that it should be transcodable to, say mp3, without really sacrifying quality.
These things are all a matter of taste and personal context, and as we see, an application oriented naming scheme leads to these subjective ambiguities.

So I think we're better off with the quality targeting approach, but here it's harder to find names.
My suggestion in clear but rather bad sounding words:

-goodEnough (for -q 2.5)
-transparent (for -q 5)
-safetyMargin (for -q 7.5)

or in short word form:

-fine (for -q 2.5)
-transparent (for -q 5)
-overkill (for -q 7.5)

with 'overkill' back as a rough equivalent to 'safetyMargin'. The added emotionality of 'overkill' isn't bad in this context IMO. It conflicts less with 'transparent' than a 'safetyMargin'. With this quality scale I wouldn't mind using 'overkill' though 'overkill' for -q 7.5 has the disadvantage that -q 5 already is expected to be overkill for most of us.

IMO this describes well what the quality settings are about.

Sure -transparent is a claim (after all we're in a world of lossy codecs), but IMO not a bad one. In case -transparent should be found not to be transparent on a sample it is a challenge to change the -transparent internals in order to reestablish transparency.
lame3995o -Q1.7 --lowpass 17

lossyWAV 1.1.0 Development Thread.

Reply #106 –
I think:

-medium (2.5)
-normal (5)
-insanity / crazy (7+)


This way things are simple and we are down to 3 real world scenarios - modest needs, normal or don't want to be bothered knowing.. then Q5 caters for the clueless while keeping true to the goals of lossywav real transparent sound at a much reduced bitrate compared to lossless. 'Insanity' gives a real hint to people that we are in overkill bitwaste territory but nonetheless it is there as an 'end to all' setting and can be considered a true replacement for lossless coding. -Medium attempts to compromise between quality and size while protectign the user from unfavourable situation where noise will be audiable and annoying - in other words -medium is probably enough for 98 % + cases but we don't want to risk ugly situations by going too low like Q0.

Quote from: halb27 on
snip..

or in short word form:

-fine (for -q 2.5)
-transparent (for -q 5)
-overkill (for -q 7.5)



I like it . Its quite clean and concise.

lossyWAV 1.1.0 Development Thread.

Reply #107 –
What happened to this thread?

Can I "me too" give my opinion

-low = -q 1
-medium = -q 2.5
-normal = -q 5 (default)
-high = -q 7.5
-insane = -q 10

OK?
The doc should not give any quality expectation for the different -q values, instead the documentation of presets should tell the corresponding -q option.

lossyWAV 1.1.0 Development Thread.

Reply #108 –
It really seems to be boiling back down to an equivalent of -1, -2 and -3 per the original development....

How about:

-P, --portable = -q 2.5;
-N, --normal = -q 5.0 (default);
-E, --excessive = -q 7.5?
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 1.1.0 Development Thread.

Reply #109 –
you could as well still use word describing targeted use & include quality expectation as a clue in the help:

-q 2.5 = -portable (good enough)
-q 5.0 = -default (transparent)
-q 7.5 = -transcoding (safety margin)

quote jido:
-low = -q 1

... ??? sorry but it starts at -q 0, you'd rather not give your opinion if you don't use the codec    unless it's a typo indeed  just teasing I already made the same typo

lossyWAV 1.1.0 Development Thread.

Reply #110 –
lossyWAV beta 1.0.1i 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 1.1.0 Development Thread.

Reply #111 –
Probably too late, but

medium/portable
standard
extreme
insane

has a good tradition on these boards!

Cheers,
David.

lossyWAV 1.1.0 Development Thread.

Reply #112 –
Quote from: 2Bdecided on
Probably too late, but

medium/portable
standard
extreme
insane

has a good tradition on these boards!

Cheers,
David.
Never too late for you, Sir! Give me 5 mins....
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 1.1.0 Development Thread.

Reply #113 –
No Nick - calm down!

I wasn't demanding all of these - just suggesting "standard" as a posh way of saying "normal" - it sounds fractionally grander.

I like "transcoding" anyway - it makes sense. When I get chance, I'm going to run some transcoding tests on lossyWAV and see which values are really worth it - so far, we're just guessing!

Given my usual speed of getting around to things, if anyone else wants to try, please jump in!

Cheers,
David.

lossyWAV 1.1.0 Development Thread.

Reply #115 –
I'd like to see another option name changed: --lowpass.
With --lowpass everybody not intimate to the lossyWAV details has the impression that the signal is lowpassed whereas it's lossyWAV's major advantage that there's nothing in the signal path changed at all - for the sake of significantly reducing the bitrate just a little bit of quantization noise is added and controlled so that it's inaudible in the sense the quality settings tell.

The --lowpass parameter addresses the frequency limits of the FFT analyses, so a name like --FFTlimit is more appropriate.
lame3995o -Q1.7 --lowpass 17

lossyWAV 1.1.0 Development Thread.

Reply #117 –
Quote from: 2Bdecided on
... just suggesting "standard" as a posh way of saying "normal" - it sounds fractionally grander. ...

I feel like this as well, and I really like your naming schemes 'medium'/'standard'/'extreme'.
lame3995o -Q1.7 --lowpass 17

lossyWAV 1.1.0 Development Thread.

Reply #118 –
Quote from: halb27 on
I'd like to see another option name changed: --lowpass.
With --lowpass everybody not intimate to the lossyWAV detail has the impression that the signal is lowpassed whereas it's lossyWAV's major advantage that there's nothing in the signal path changed at all - for the sake of significantly reducing the bitrate just a little bit of quantization noise is added and controlled so that it's inaudible in the sense the quality settings tell.

The --lowpass parameter addresses the frequency limits of the FFT analyses, so a name like --FFTlimit is more appropriate.
I would prefer --hflimit or --upperlimit or something like that - --FFTlimit is a bit ambiguous....
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 1.1.0 Development Thread.

Reply #119 –
Quote from: halb27 on
I'd like to see another option name changed: --lowpass.
With --lowpass everybody not intimate to the lossyWAV detail has the impression that the signal is lowpassed whereas it's lossyWAV's major advantage that there's nothing in the signal path changed at all - for the sake of significantly reducing the bitrate just a little bit of quantization noise is added and controlled so that it's inaudible in the sense the quality settings tell.

The -lowpass parameter addresses the frequency limits of the FFT analyses, so a name like -FFTlimit is more appropriate.
I agree. I'd suggest "analysislimit" but it's longer to type!

Quote from: Nick.C on
I would prefer --hflimit or --upperlimit or something like that - --FFTlimit is a bit ambiguous....
Did you leave "lower limit" internally within the code, or is it gone? Not that I need it - just curious.

EDIT:

Maybe just -limit with the explanation "frequencies higher than this are not analysed"

... and maybe extended help "lossyWAV adds white noise based on analysis of the signals _below_ this frequency; if signals above this frequency are at an even lower level, they they can be swamped by the added noise. This is usually inaudible, but the behaviour can be changed by specifying a higher -limit. For many audio signals, there is little content at very high frequencies, and forcing lossyWAV to keep the noise lower than the content at these frequencies can increase the bitrate dramatically"

or something!

Cheers,
David.

lossyWAV 1.1.0 Development Thread.

Reply #120 –
Quote from: 2Bdecided on
No Nick - calm down!
<deep breath>
Quote from: 2Bdecided on
I wasn't demanding all of these - just suggesting "standard" as a posh way of saying "normal" - it sounds fractionally grander.
Agreed.
Quote from: 2Bdecided on
I like "transcoding" anyway - it makes sense. When I get chance, I'm going to run some transcoding tests on lossyWAV and see which values are really worth it - so far, we're just guessing!
Does --transcoding not imply some sort of internal transcoding process rather than the-output-is-considered-to-be-good-enough-for-transcoding?

So, in the spirit of keeping to tradition  :

--insane = -q 10;
--extreme = -q 7.5;
--standard = -q 5.0;
--portable = -q 2.5.

Quote from: 2Bdecided on
I agree. I'd suggest "analysislimit" but it's longer to type!
-A, --analysislimit or -U, --upperlimit?

Quote from: 2Bdecided on
Did you leave "lower limit" internally within the code, or is it gone? Not that I need it - just curious.
It's gone - I could relatively easily re-instate it (20<=n<=1378.125) --lowerlimit.
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 1.1.0 Development Thread.

Reply #121 –
Quote from: Nick.C on
I would prefer --hflimit or --upperlimit or something like that - --FFTlimit is a bit ambiguous....

I would avoid a word that can be associated as a synonym for 'lowpass'.
My first thought was 'analysislimit' like 2Bdecided suggested, but for the sake of shortness I changed it to FFTlimit. 'analysislimit' is more precise of course, so why not use it.
lame3995o -Q1.7 --lowpass 17

lossyWAV 1.1.0 Development Thread.

Reply #122 –
see my edited post above (I can't keep up!)

lossyWAV 1.1.0 Development Thread.

Reply #123 –
Quote from: Nick.C on
...
--insane = -q 10;
--extreme = -q 7.5;
--standard = -q 5.0;
--portable = -q 2.5.

I like it - though I'd prefer 'medium' as all the other names refer to quality and not application. But it's not really important.


Quote from: 2Bdecided on
see my edited post above (I can't keep up!)

--limit is ok IMO though --analysislimit is clearer.
lame3995o -Q1.7 --lowpass 17

lossyWAV 1.1.0 Development Thread.

Reply #124 –
It would be necessary to add an lowpass limit for portable preset.
Wavpack -hh or TAK -pMax
OggVorbis aoTuVb6.03 -q 4