Winamp 5.6 Rating Implementation Discussion (split from the 5.6 release thread)

[DJ Egg]

@Fla_Panther

We already do use the same method as Windows / WMP.
It's MM which doesn't.
In the next release update, we'll be re-mapping MM's rating scale to the Winamp/WMP method.

Winamp rating scale is:
0 = 0
1-63 = 1
64-127 = 2
128-195 = 3
196-254 = 4
255 = 5

Winamp & Windows/WMP write the values as:
0=0
1=1
2=64
3=128
4=196
5=255

MM is doing something funky (see the link in my first reply).
But no worries, in the next release we'll have it mapped to read MM's rating the same as ours.

If MM ever decides in the future to change their method to match ours, then hopefully they will also change their email identifier in the popm frame (because we'll only apply the adjustment if it's: no@email).

[Edit]
Our rating scale has since been shown to be wrong. See page 3 for details/info/solution/etc.
Note, we will not be using the above method, i.e. we won't be reading MM's no@email from POPM, so please ignore the above.

[Fla_Panther]

Originally Posted by DJ Egg

We already do use the same method as Windows / WMP.
It's MM which doesn't.

Alright, fair enough.

That still leaves me stuck trying to get my music into WinAmp until the next release. Any ETA on when that might be?

I tried mp3tag and it doesn't even show my ANY ratings, much less what program rated the songs. I guess I'll PM Sinatra and maybe he can explain to me how to get the program to show me that stuff. Even then I'm not sure how I'll be able to get the ratings converted.

[MrSinatra]

Originally Posted by Fla_Panther

Alright, fair enough.

its not true! see post above.

Originally Posted by Fla_Panther

That still leaves me stuck trying to get my music into WinAmp until the next release. Any ETA on when that might be?

i doubt they will give you an ETA.

but if they just do what windows does, which they mistakenly claim [honestly, i'm sure] is what they already do, you won't have a problem.

Originally Posted by Fla_Panther

I tried mp3tag and it doesn't even show my ANY ratings, much less what program rated the songs. I guess I'll PM Sinatra and maybe he can explain to me how to get the program to show me that stuff. Even then I'm not sure how I'll be able to get the ratings converted.

see my earlier link to mp3tag. they do the same damn dumb thing that is proposed here, (kinda).

if i had a POPM that said this:

Me@myemail.com | 255 | 0

then i would see that raw data, AS I SHOULD, in a column called "Popularimeter"

%popularimeter%

but for some damn fool reason Florian will take all the POPMs that have these specific ID strings / email addys:

rating@winamp.com | 255 | 0
no@email | 225 | 0
Windows Media Player 9 Series | 255 | 0

and interpret it for you, as just "5" in mp3tag

the columns are based on:

%rating winamp%
%rating wmp%
%rating mm%

and can be seen in the "extended fields" when creating a new column.

this is beyond dumb, as that info is what i can see in the app already. the whole point of mp3tag is to allow you to see the RAW data and manipulate it.

i proposed that he should keep it all that way, but just allow popularimeter:

%popularimeter%

...to not be pointlessly disabled, but simultaneously able to display the raw data, just like it used to.

why he refuses to even post on the subject is one of lifes great mysteries, and disappointments, to me.

in any case, using the id string for determining scales, or interpreting how to display data, whatever, is just lame. the point of the string is to allow differernt users to have different ratings on one file. thats ALL an app should use it for, period.

[ryerman]

@ DJ Egg

Originally Posted by DJ Egg

......
WMP & Winamp rating scale is:
0 = 0
1-63 = 1
64-127 = 2
128-195 = 3
196-254 = 4
255 = 5
.....

This seems to imply that Winamp currently deciphers (parses?) the ID3 POPM frame to extract a rating value that is converted to its own internal star-rating system, according to this table.

The table seems to indicate that Winamp should convert the hexadecimal rating value 81 (decimal equivalent 129) to a 3 star rating but my experience is that only 2 stars will be shown.

I used the "right click, rate items" method to give a file a 3 star rating. Winamp wrote 80 (hex).
I used a hex editor and changed the appropriate byte from 80 to 81. Winamp then showed 2 stars.

Something seems inconsistent. An increase in numerical rating value should not result in a decrease in rating stars.

[MrSinatra]

Originally Posted by DJ Egg

WMP & Winamp rating scale is:
0 = 0
1-63 = 1
64-127 = 2
128-195 = 3
196-254 = 4
255 = 5

for reference below, but again, that is NOT the windows [explorer] scale. and its WMP only insofar as granularity is concerned, but strictly speaking 254 would not = 4, but something closer to 4.9, which is then how WMP would display the stars.

Originally Posted by ryerman

@ DJ Egg

This seems to imply that Winamp currently deciphers (parses?) the ID3 POPM frame to extract a rating value that is converted to its own internal star-rating system, according to this table.

i'm with you so far i think, winamp converts the # data to a star, yes.

Originally Posted by ryerman

The table seems to indicate that Winamp should convert the hexadecimal rating value 81 (decimal equivalent 129) to a 3 star rating but my experience is that only 2 stars will be shown.

i am not sure why you are involving hex values?

but in any case, i took a file, and gave it a 129 POPM value, and winamp reported it as 2 stars! you are correct! so somehow, Egg/winamp has it wrong.

i believe the reason is that this gets back to winamp using the wrong ranges, and then compounds it with complicated (and unncessary) math to figure out what should be a straight, one for one, table lookup. this value = this star.

winamp definitely writes and reads POPM=128 as THREE stars, so there is no question that 129 being reported as TWO stars is WRONG.

Originally Posted by ryerman

I used the "right click, rate items" method to give a file a 3 star rating. Winamp wrote 80 (hex).
I used a hex editor and changed the appropriate byte from 80 to 81. Winamp then showed 2 stars.

Something seems inconsistent. An increase in numerical rating value should not result in a decrease in rating stars.

agreed, although it would be better if we could just stick to POPM #s and leave hex out, at least for me, b/c i think hex could confuse the issue.

bottom line, winamp is currently inconsistent even within just ITSELF, and needs to follow this chart instead, which is straight forward and will fix all the issues, esp if done as a straight table lookup, as opposed to an unnecessary calculation:

>>
224-255 = 5 stars when READ with windows explorer, writes 255
160-223 = 4 stars when READ with windows explorer, writes 196
096-159 = 3 stars when READ with windows explorer, writes 128
032-095 = 2 stars when READ with windows explorer, writes 64
001-031 = 1 stars when READ with windows explorer, writes 1
>>

[MrSinatra]

Originally Posted by DJ Egg

@Fla_Panther

We already do use the same method as Windows / WMP.

no, you don't. that is absolutely clear.

if you did, then you would see his MM file as 5 stars, which windows does, but winamp does not.

thats the whole reason he posted to begin with.

the reason this is true is b/c winamp, for some unknown reason to me, currently uses different ranges than windows [explorer] does to mean 'this # = this star.'

Originally Posted by DJ Egg

It's MM which doesn't.

MM apparently writes values that windows explorer (or winamp) does not. we know this b/c his file uses 252 for 5 stars, while windows explorer / winamp write 255 for that.

but who cares what MM does? i mean, according to that thread Egg posted, MM has had at least two different scales, both presumably using the same id string. you can't know which scale a user will bring to you, if doing special case handling based on the faulty assumption that the string will mean something concrete.

all that matters to MM users trying to port into winamp, is that whatever MM writes, it falls into the ranges that windows explorer uses and that winamp should be compliant with.

i'm not going to test MM, but i'm 99% sure that whatever it writes, its complies with the ranges that windows explorer READS.

Originally Posted by DJ Egg

In the next release update, we'll be re-mapping MM's rating scale to the Winamp/WMP method.

WMP & Winamp rating scale is:
0 = 0
1-63 = 1
64-127 = 2
128-195 = 3
196-254 = 4
255 = 5

that is absolutely FALSE.

it MAY represent how winamp reads POPM data ranges, (i haven't tested it, but i don't doubt you), but windows/WMP uses this scale:

>>
224-255 = 5 stars when READ with windows explorer, writes 255
160-223 = 4 stars when READ with windows explorer, writes 196
096-159 = 3 stars when READ with windows explorer, writes 128
032-095 = 2 stars when READ with windows explorer, writes 64
001-031 = 1 stars when READ with windows explorer, writes 1
>>

keep in mind that WMP supports granularity, but lets not confuse the topic, since we are really only concerned with how winamp currently displays stars.

Originally Posted by DJ Egg

Winamp & Windows/WMP write the values as:
0=0
1=1
2=64
3=128
4=196
5=255

thats true, but again keep in mind that WMP supports granularity, so what it writes goes beyond just those "whole star" values.

Originally Posted by DJ Egg

MM is doing something funky (see the link in my first reply).
But no worries, in the next release we'll have it mapped to read MM's rating the same as ours.

If MM ever decides in the future to change their method to match ours, then hopefully they will also change their email identifier in the popm frame (because we'll only apply the adjustment if it's: no@email).

winamp should not be concerned with how MM reads or writes POPM values, and as you alude yourself, the scale could change yet again and still keep the same string, so its folly to follow that. i think at this point, i've made clear why, but if winamp just did what windows does, MM users would be fine.

[MrSinatra]

ok. i am finally ready to post about this, more or less.

after extensive testing on my end and searching lots of crap tag editors, i finally found something tedious to use, but workable to PROVE my points. i am hungover, so i am cranky, so maybe take what i am about to say with a grain of salt, but DrO and DJ Egg, i love you guys, and i respect you both A LOT but i have to say it angers me when you both dis me after everything i have contributed to your guys cause. i'm not saying you personally disrespected me, but i am saying you professionally disrespected me, imo. and it just bothers me b/c i know you guys know i am good at what i do, and that i know ratings better than you both put together. no need to dwell on this, but honestly i feel it needed said. just give me a little credit, i'm not a noob.

ok... now, a lot of what has been said is CATEGORICALLY FALSE. so i hardly know where to start, but i will go back and try to reply to each post.

[DJ Egg]

Right... so we now need to find out how/where we got our 0-255 ratings scale (0, 1-63, 64-127, 128-195, 196-254, 255) from originally (ie. years ago), and why it was wrong....

Has WMP changed it since we implemented it, or has it always used that scale (0, 1-31, 32-95, 96-159, 160-223, 224-255)?

ps. calm down :-p :-)

[MrSinatra]

Originally Posted by DJ Egg

Right... so we now need to find out how/where we got our 0-255 ratings scale (0, 1-63, 64-127, 128-195, 196-254, 255) from originally (ie. years ago), and why it was wrong....

ok, like i've said a few times now, WMP supports granularity in stars. so its going to be different in a way from windows explorer. but they don't contradict each other, and thats why if winamp would just copy the way windows explorer READS ratings, as opposed to WMP (which is probably what you used years ago, and years ago you all also decided 4.9 WMP stars, aka 254, would be 4 winamp stars), EVERYONE would be happy, including MM users!

see my chart!

Originally Posted by DJ Egg

Has WMP changed it since we implemented it, or has it always used that scale (0, 1-31, 32-95, 96-159, 160-223, 224-255)?

don't confuse things!

that scale is what WINDOWS EXPLORER uses, not WMP!

afaik, windows has used that same scale for years and years.

WMP probably uses a scale closer to what you posted earlier, but like i said in the previous post, 254 doesn't = 4, it equals something closer to 4.9, thats the granularity!

the good news for you is you do NOT need to change what values winamp WRITES, you only need to change what winamp READS, meaning it should interpret the values on my chart to mean such and such a star.

you also need to do a straight table lookup, and give up the unnecessary math/calculations, b/c thats probably why 129 is mistakenly giving us two stars, BAD MATH DESIGN!

Originally Posted by DJ Egg

ps. calm down :-p :-)

i love you egg, and i'm calm. i'm just trying to make my case vigorously, and also respect all the previous posts/users by addressing the points in them. methodical.

i hope by now i have earned faith.

[DJ Egg]

http://msdn.microsoft.com/en-us/libr...=vs.85%29.aspx

We're looking into it...

What a carry on! :-(


Also note, for now, we're just concerned with reading the correct 1-5 rating, not fractions (because support for that isn't implemented in Winamp).

I think I remember a discussion on here a few years ago (maybe Will was involved?) where we were debating whether to round up the rating (on read) to the nearest whole integer, e.g. if we see a rating of 4.6-4.9 (~230-254?), as written by some external app/editor, should it be read in Winamp as a 4 or 5 star rating, and we decided on the former....

I do understand your concerns/reasoning/etc on those issues, but it's not relevant, for now....

Let's just get the 1-5 ratings correct first.

[EDIT: On reflection, the above was actually a result of erroneous implementation, related to both the scale we were using and the reading of it - all fixed for 5.64]

[MrSinatra]

Originally Posted by DJ Egg

http://msdn.microsoft.com/en-us/libr...=vs.85%29.aspx

We're looking into it...

What a carry on! :-(

that link imo is for the WMA format, not id3/POPM.

me carrying on or the issue itself? i admit, i'm a bit over the top, but i am well intentioned.

i can tell you that the chart i gave, i personally tested each value.

224-255 = 5 stars when READ with windows explorer, writes 255
160-223 = 4 stars when READ with windows explorer, writes 196
096-159 = 3 stars when READ with windows explorer, writes 128
032-095 = 2 stars when READ with windows explorer, writes 64
001-031 = 1 stars when READ with windows explorer, writes 1

if you would implement that chart for POPM, for reading since you already write it that way, and cut out the math/calculations you do and just make it a straight table lookup, imo, you'd be golden, for all users, including MM ones. (and of course drop the useless, unnecessary email id / special case handling)

[MrSinatra]

Originally Posted by DJ Egg

Also note, for now, we're just concerned with reading the correct 1-5 rating, not fractions (because support for that isn't implemented in Winamp).

I KNOW!!! I KNOW!!! I KNOW!!!

that is EXACTLY what i have been addressing! its not ME who has confused the issue with WMP and its ranges or handling or granularity, but rather WINAMP (and its devs) who have done so!

EVERYTHING i have said has been to get the right 1-5 stars displayed in winamp!

fractions only apply to display but explain why values other than 1,64,128,196,255 exist, thats all.

Originally Posted by DJ Egg

I think I remember a discussion on here a few years ago (maybe Will was involved?) where we were debating whether to round up the rating (on read) to the nearest whole integer, e.g. if we see a rating of 4.6-4.9 (~230-254?), as written by some external app/editor, should it be read in Winamp as a 4 or 5 star rating, and we decided on the former....

i don't know, i don't think i was involved, but again, you are missing the point imo.

just follow my chart! my chart is what WINDOWS EXPLORER does! thats what you want to use for READING THE RANGES. what you WRITE is already correct. but the way winamp currently READS the ranges is INCORRECT, ergo why MM users have the problem.

i really don't see whats confusing to you about that? or if its not confusing, what it is about that, that you don't want to do?

Originally Posted by DJ Egg

I do understand your concerns/reasoning/etc on those issues, but it's not relevant, for now....

Let's just get the 1-5 ratings correct first.

with all due respect, i don't think you do understand what i am saying. i am trying to get the 1-5 part correct, just as you want me to. but you are not understanding me!

what i am saying is incredibly relevant! 252 is seen by windows explorer and MM as 5 stars, its ONLY winamp that sees it as 4!

follow my chart, and winamp will get it right!

[DJ Egg]

Yeah, it's a 0-100 scale there, but (except for the "1") it matches your values when converted to the 0-255 scale (*/2.55 and rounded-up to nearest/highest integer). I just needed to see some official documentation :-)



ps. "What a carry on!" is a 'jokey' saying (of British origin), which roughly translates as "what a fuss/commotion/rumpus/shemozzle/etc"

[Aminifu]

Originally Posted by DJ Egg

Yeah, it's a 0-100 scale there, but (except for the "1") it matches your values when converted to the 0-255 scale (*/2.55 and rounded-up to nearest/highest integer). I just needed to see some official documentation :-)

Will you post a link to any official Microsoft documentation you find?

I've been following the discussion of changes for equating the 0-255 POPM ratings to the 0-5 star ratings to be displayed in the next version of Winamp. It's interesting that, so far, what Microsoft does is not the de-facto standard agreed on throughout the industry. Digital media is here to stay and it is pass time some standards body or group of vendors resolves this and other tagging issues for all formats that support tagging, imo.

It logical to assume equal degrees or steps of granularity between each full star rating, but that would require 51.2 degrees or steps (256/5) for the current POPM range. That is an awkward number to deal with, so the compromise that Windows Explorer uses (2 ranges of granularity), seems reasonable (imo), and is easy to compute:

(x/64) + 1 = number of stars, where x is the POPM rating from 1 to 255 with normal rounding of the result.

This results in 32 degrees or steps between 0 to 1 and 4 to 5 stars and 64 degrees or steps between 1 to 2, 2 to 3, and 3 to 4 stars, as determined by MrSinatra's research of Windows Explorer.

Thank you, MrSinatra for helping to resolve the inconsistent handling that Winamp currently does.

[MrSinatra]

Originally Posted by Aminifu

Will you post a link to any official Microsoft documentation you find?

Originally Posted by DJ Egg

http://msdn.microsoft.com/en-us/libr...=vs.85%29.aspx

he did post that, but i assume you saw that?

i think its something microsoft just kind of invented out of thin air.

Originally Posted by Aminifu

I've been following the discussion of changes for equating the 0-255 POPM ratings to the 0-5 star ratings to be displayed in the next version of Winamp. It's interesting that, so far, what Microsoft does is not the de-facto standard agreed on throughout the industry. Digital media is here to stay and it is pass time some standards body or group of vendors resolves this and other tagging issues for all formats that support tagging, imo.

so far, imo, just about everything tagging wise is set by de facto standard, and the industry, esp apple, ignores whatever published specs there are at will.

but even something like MM is guilty. why on earth would they choose 252 to equal 5 stars? its just baffling.

to me, the sensible thing to have done would have been to set POPM and vorbis RATING with a 0-100 data range in a spec, but even had that happened, one wonders would it have been ignored? look at the abuse of the TPE2 frame for such wanton disregard.

in any case, the only option that makes sense here, is to follow what microsoft does, which thankfully in this case, isn't totally outrageous in its implementation.

Originally Posted by Aminifu

It logical to assume equal degrees or steps of granularity between each full star rating, but that would require 51.2 degrees or steps (256/5) for the current POPM range. That is an awkward number to deal with, so the compromise that Windows Explorer uses (2 ranges of granularity), seems reasonable (imo), and is easy to compute:

(x/64) + 1 = number of stars, where x is the POPM rating from 1 to 255 with normal rounding of the result.

This results in 32 degrees or steps between 0 to 1 and 4 to 5 stars and 64 degrees or steps between 1 to 2, 2 to 3, and 3 to 4 stars, as determined by MrSinatra's research of Windows Explorer.

i don't really know, but i don't think the spec designers counted 0.000 when they designed the range, since 0 is essentially the same as unrated; one would not be able to tell the difference in a star rating.

instead, they called it 1-255 or 51 equal steps of 5, and the odd number 51 allowed to set an exact mid-point, (of 26th place).

in such a scheme, it would look like this:

0=0=unrated=irrelevant

001-050=granular points above 0, but below 1, and 26=halfway, b/c 0 is bottom, 51 is top

051=1

052-101=granular points above 1, but below 2, and 77=halfway

102=2

103-152=granular points above 2, but below 3, and 128=halfway

153=3

154-203=granular points above 3, but below 4, and 179=halfway

204=4

205-254=granular points above 4, but below 5, and 230=halfway

255=5

and just to be clear, 26 is halfway b/c there are 25 granular points to the left and right of it, (and in this system, you do count the 0-1 space, as well as the 50-51 as points).

also notice how here, 128 is actually 2.5 stars, not 3 as winamp/windows explorer deem it.

thats all just my guesswork, and i would never design such a stupid range. but interestingly, its probably closer to what WMP uses b/c ONLY 255 = 5, everything else is less than 5 to WMP. i think thats what screwed up winamp in the first place, trying to copy WMP instead of explorer.

Originally Posted by Aminifu

Thank you, MrSinatra for helping to resolve the inconsistent handling that Winamp currently does.

i appreciate that, thx. Ryerman also had a great spot (thats currently unexplained) in that 129 gives u less stars than 128 (in winamp), but i think its the WMP reason.

the code that DrO did based on my chart should eliminate all such issues and comply with windows explorer, and likely solve any issues for MM users or others.

[DrO]

code:
if (rating >= 224 && rating <= 255)

	rating = 5;

else if (rating >= 160 && rating <= 223)

	rating = 4;

else if (rating >= 96 && rating <= 159)

	rating = 3;

else if (rating >= 32 && rating <= 95)

	rating = 2;

else if (rating >= 1 && rating <= 31)

	rating = 1;

else

	rating = 0;

that is the final code change on this (ever!). everything else from the MM specific changes to including trying to deal with 0-10 and 0-5 is gone.

and i am pissed off about this as we said how we were doing things when it all went in and people involved in this thread agreed that it was correct (and there is no need for _anyone_ to reply to that comment, ok!).

either way, it's done. it's now finished and i don't want to see war & peace like tomes of replies in response to this.

[MrSinatra]

Originally Posted by DrO

if (rating >= 224 && rating <= 255)
rating = 5;
else if (rating >= 160 && rating <= 223)
rating = 4;
else if (rating >= 96 && rating <= 159)
rating = 3;
else if (rating >= 32 && rating <= 95)
rating = 2;
else if (rating >= 1 && rating <= 31)
rating = 1;
else
rating = 0;

that is the final code change on this (ever!). everything else from the MM specific changes to including trying to deal with 0-10 and 0-5 is gone.

what i see there looks right to me, as far as POPM is concerned. thats my chart, which is based on windows explorer. it should work just fine for MM users, and really, all users.

not sure what the 0-10 and 0-5 stuff refers to? POPM? or do you mean the vorbis stuff?

no skin off my nose if its a true 0-100 scale for vorbis now, (thats the scale i use, and the de facto marketplace fave), BUT i thought the consensus was that the special handling for 2-5 should remain for those who used a 0-5 vorbis scale?

Originally Posted by DrO

and i am pissed off about this as we said how we were doing things when it all went in and people involved in this thread agreed that it was correct (and there is no need for _anyone_ to reply to that comment, ok!).

either way, it's done. it's now finished and i don't want to see war & peace like tomes of replies in response to this.

i've no idea what you're angry about, but honestly i feel like no one should be angry at me. i tried to head off people wasting their time. and i am not paid by anyone, yet i have put in lots of time on this, did the testing, and got you all the best possible solution, which i am grateful you have now implemented.

is it too much to expect some gratitude? some thanks? some admission that i was correct?

[DJ Egg]

@MrSinatra
There's no confusion.
Now we know the correct values, we will be following them.
Blame whoever/wherever we got the wrong values from originally.

[Edit]

Umm, yeah, what DrO said :-)


So basically, this latest change will be global, for all POPM frames, and will also fix the MM issue (ie. we'll now also read MM's ratings correctly). Yay!
And to clarify, the writing code hasn't been changed, just the reading code.

[Aminifu]

Hi MrSinatra,

I was following the whole discussion while I was at work. DJ Egg was talking about official documentation to support your suggestion. Since he and DrO decided to implement it, maybe he will stop looking. But if not, it would be good to see what he finds.

I read ryerman's comments. That's why I said the current inconsistent handling that Winamp does.

Are you an Eagles fan? I'm a homer, so the Bears are my favorite. Anyway, time for bed now. I have 1 more 12 hour shift to do tonight.

[MrSinatra]

Originally Posted by Aminifu

Hi MrSinatra,

I was following the whole discussion while I was at work. DJ Egg was talking about official documentation to support your suggestion. Since he and DrO decided to implement it, maybe he will stop looking. But if not, it would be good to see what he finds.

i agree, i just doubt he'll find more. he seemed to say that what he did find, that link, lent itself to the ranges i posted re: windows explorer. my guess is they are all sick of the issue, and me, and have moved on. c'est la vie.

Originally Posted by Aminifu

I read ryerman's comments. That's why I said the current inconsistent handling that Winamp does.

right, i just wanted to get him his props. i hadn't tested winamp yet, and so it was good to find that contradiction, b/c it proved that whatever winamp was doing, it was flawed.

Originally Posted by Aminifu

Are you an Eagles fan? I'm a homer, so the Bears are my favorite. Anyway, time for bed now. I have 1 more 12 hour shift to do tonight.

regrettably, yes. i am a huge Nick Foles fan, and was so even before the draft, so i have hope for the future. seeing the end of Reid and the felon Vick gives me joy.

[DJ Egg]

Ok, I've split some posts into another thread, and locked it. There's possibly more (e.g. some of mine, lol), but that will do for now....

Please play nice and keep things polite and on topic.

The ratings implementation was admittedly a mess, mostly, but has now been fixed for the next release....
except for the unfortunate issue with reading the values 2,3,4,5 in flac/vorbis tags as actual ratings, as previously discussed.

Thank you to everyone who helped resolve the overall problem :-)

[MrSinatra]

Egg,

is it planned to add the %rate% atom with a scale of 0-100 for mp4/m4a support? that would be excellent. at that point, all the major formats that can support ratings in tags, would.

[DJ Egg]

Maybe

[MrSinatra]

Egg, DrO, devs etc...

could someone please expose in an ini file the ability to specify the email string winamp uses for POPM? (only code for id3v2.x handling needs modified)

right now winamp uses:

rating@winamp.com

...by default for all POPM frames it writes fresh. i would like it to use my email addy instead. (if POPM has an existing txt ID value / email addy, winamp currently leaves it in place, as it should)

in addition, it would be great if you could also add a switch to the ini that would allow winamp to overwrite any existing email addy with the default addy or the ini specified one, but i'd understand if you didn't want to expose that in the ini. but please at least do the first bit!

[DrO]

not going to be done by me and i personally disagree with allowing it to be customised based on how all other programs do things by using their own string and it just adds complexity to the implementation which is not needed by the majority using ratings.

-daz

[MrSinatra]

hmm, i thought winamp was about allowing customization, not cloning itunes or other apps? am i wrong?

seems to me a lot of things are already done differently than by other programs, that the official prefs allow for that even more, and for the less "popular" things exposing stuff in the ini is not unheard of, (meaning, you guys do that quite a bit already for other stuff, which i think is great ).

i'm not sure what complexity is even added? this would only be invoked by those modifying the ini, so existing handling would stay as is, unless replaced via ini invocation by the handling i am asking for. and i agree the majority aren't concerned (yet) with this, but thats why i asked for ini invoked handling, to service the minority who do care, which i hope winamp still cares about.

EDIT:

i should probably point out for anyone reading this, that the spec says the email addy ID string in the POPM frame is to identify the USER not the app or anything else. that some apps use their own string mandatorily by default, is really nothing more than just an abuse of the spec.

[DrO]

fine. it's done. you'll have to use rating_email=<your_email_address> in the Nullsoft MPEG Decoder section of winamp.ini when it's not running. that will then prefer your "email" over using "rating@winamp.com" for reading and writing of the rating but will revert to that if your "email" is not specified or not found on reading (so it'll allow migration when re-rating the values, etc).

i still heavily disagree about this so if it screws up your files then i accept no blame.

as for the 'spec', no one else is doing that by default and i would interpret the 'user' as the program doing it as it otherwise requires _every_ program which does rating handling to ask for an email address to use on install, etc.

[DrO]

Originally Posted by DrO

you'll have to use rating_email=<your_email_address> in the Nullsoft MPEG Decoder section of winamp.ini when it's not running.

scratch that, you'll get a preference option instead as per the attachment.

-daz

[MrSinatra]

thanks again. please don't get flustered by the length of this post, or the contents of it, its just me being methodical.

Originally Posted by DrO

fine. it's done. you'll have to use rating_email=<your_email_address> in the Nullsoft MPEG Decoder section of winamp.ini when it's not running. that will then prefer your "email" over using "rating@winamp.com" for reading and writing of the rating but will revert to that if your "email" is not specified or not found on reading (so it'll allow migration when re-rating the values, etc).

i still heavily disagree about this so if it screws up your files then i accept no blame.

can you explain the above a bit? i wanted to ask sooner but i just didn't have the time.

--

am i correct to assume that even if multiple POPM frames are present, that winamp will still only read one no matter what? if so, won't winamp always read whatever is presented by the POPM frame it "sees" regardless of the email value in the pref or the file?

--

also, i think a further on/off switch/checkbox is needed, i'll explain...

if a user changes the default value of rating@winamp.com to something else (or even if they don't), then there are two things they may legitimately want:

1. they may want the pref's ID value to be written in ALL cases,
or
2. they may want the pref's ID value to be written ONLY when an existing POPM ID value is NOT already present. (ie. a totally new rating for that previously unrated file)

as of 5.63, winamp properly respects existing ID values when making rating/star updates, but writes its own value when none is [already] present. this is proper and good behavior, since winamp can not and should not assume existing user IDs are not there on purpose. this should remain the default winamp behavior.

having said that, a user might want winamp to overwrite existing IDs, (as i do), when they try to, as you say, migrate. right now, i have thousands of rating@winamp.com IDs (as well as some random others) i want to migrate to my own email. a checkbox would allow me to turn on that ability, when i need it, and then turn it off when i'm done.

i hope thats not a problem?

--

also, i assume winamp will not allow a user to leave this pref value totally blank? i think some kind of entry should be mandatory, and if they do try to make it blank, it should revert to rating@winamp.com

Originally Posted by DrO

as for the 'spec', no one else is doing that by default and i would interpret the 'user' as the program doing it as it otherwise requires _every_ program which does rating handling to ask for an email address to use on install, etc.

well actually, i'm not sure that "no one else is doing that" but even if true, we should not be lemmings. winamp has the opportunity to leverage power out of the spec that other big apps, i agree, aren't. (i agree with your pragmatism, you are conforming to the de facto stance of these "big" apps, i'm just disagreeing that its necessary to). also, i think it is clear that the spec is refering to people as users, not apps, esp since it wants an email to differentiate ratings within the same file.

what every program should do, is what you are allowing winamp to do. let the user decide what to be written when writing ratings.

what would be fantastic, would be the ability of apps, not just winamp, to recognize multiple POPM frames. consider this scenario:

a family has many computers and devices, and music centrally stored on a server or NAS. each user has their own winamp, maybe even multiple winamps. (or, maybe one computer with multiple windows logins/users, each with a user specific winamp install).

now lets say each user put their email in the pref as you have now implemented it. if in addition, multiple POPM frames WERE supported, then each file could have dozens of user specific ratings in it, AND each winamp install could read and write JUST that users POPM frame, based on your new pref!

that would eliminate a lot of your objections to ratings in tags, b/c they would then be user specific. and it would mean that what i said earlier wouldn't matter, meaning that in such a scenario as this, winamp WOULD and SHOULD read and write ONLY the user ID value in your new pref.

i know supporting multiple id3 frames is beyond anything you want to do here and now, but surely you see where i'm coming from? and this new pref is helpful for taking steps to that direction.

and such additions would not be a burden, b/c most users could just leave the defaults in place and be happy with it. ergo, the app would not need to ask for that information. (i should point out, that winamp DOES ask for an email when installing, it just doesn't require it. but it could optionally use that email for the pref)

one last thing, is that playcounts are also supported by POPM, and could also be user specific as per the above. that way, they wouldn't be lost when the DB borks as ratings once were, (assuming the user writes ratings to tags). making playcounts user specific via POPM is probably more desirable then supporting PCNT, but both could be done, and PCNT could just represent the "global" amount.

not requesting any of that here, just saying the spec lends itself to being leveraged IF apps take advantage of it and implement it the right way.

EDIT: regardless of rating updates or this new pref, etc, winamp should not overwrite any existing POPM playcount values.

Originally Posted by DrO

scratch that, you'll get a preference option instead as per the attachment.

-daz

once again, thanks! the screenshot looks great and i love that the default value is displayed.

[Fla_Panther]

Hi all. These rating issues have still been floating through my mind for the last few months but I've had other things on my plate. Now that I've had some time to come back Iv'e caught up on the thread and Mr. Sinatra's posts took the same path as my thoughts. Unfortunately DrO skipped a significant section as the overall post was tl;dnr. If I may, I'd like to bring back this one section:

Originally Posted by MrSinatra

a family has many computers and devices, and music centrally stored on a server or NAS. each user has their own winamp, maybe even multiple winamps. (or, maybe one computer with multiple windows logins/users, each with a user specific winamp install).

now lets say each user put their email in the pref as you have now implemented it. if in addition, multiple POPM frames WERE supported, then each file could have dozens of user specific ratings in it, AND each winamp install could read and write JUST that users POPM frame, based on your new pref!

that would eliminate a lot of your objections to ratings in tags, b/c they would then be user specific. and it would mean that what i said earlier wouldn't matter, meaning that in such a scenario as this, winamp WOULD and SHOULD read and write ONLY the user ID value in your new pref.

i know supporting multiple id3 frames is beyond anything you want to do here and now, but surely you see where i'm coming from? and this new pref is helpful for taking steps to that direction.

and such additions would not be a burden, b/c most users could just leave the defaults in place and be happy with it. ergo, the app would not need to ask for that information. (i should point out, that winamp DOES ask for an email when installing, it just doesn't require it. but it could optionally use that email for the pref)

.... This is exactly what I've been thinking. All the media players have been misusing the tag (not throwing stones, bear with me). From what I've read, one reason Apple has a database separate from the files is to do exactly this - handle ratings for multiple users. But this could be done in the file without any DB needed.

In my case, as someone who makes mixes now and then I'd love to be able to have a music library program that will let me tag songs with separate ratings - one for me as an individual, and another for me as a DJ - because I would rate songs differently for myself versus for a mix set. Or for a professional DJ, it could be handy to have different ratings for different types of gigs (wedding, bar mitzvah, birthday, etc). This is no different than allowing a family of 4 to have their own ratings. And yes, I store my files on a NAS, so I want to be able access them from any PC in the house and see the ratings I want.

Granted, giving Winamp a drop-down menu to select which user's ratings are shown is far off in the future. But the ability to tag with a truly user-generated ID is the future, IMO.

[MrSinatra]

Originally Posted by Fla_Panther

Granted, giving Winamp a drop-down menu to select which user's ratings are shown is far off in the future. But the ability to tag with a truly user-generated ID is the future, IMO.

the ability to specify your email addy for POPM is now in the beta. but that's it. no multiple POPM frames or anything else.

the dropdown menu is not the hard part for the devs. implementing multiple POPMs and then creating a UI in prefs to control how it would work is the hard part, (although, clearly doable, just a matter of them wanting to do it, and then taking the time to do it). not to mention making the ML display/act differently depending how on the fly changes to the values the user sets to those prefs.

here is what would need to be done:

1. support multiple POPM frames.
2. exposing in the prefs the ability for the user to delineate what email id string is written to POPM. (default: rating@winamp.com ) THIS IS DONE.

note: it would be nice to be able to store multiple emails (or other txt ID strings) however, and pick which one you want via a dropdown box or similar.

3. supplying a toggle to overwrite or respect existing id txt strings, (default: respect)

this would determine if new POPM frames would be written or not. if set to "respect" then when a rating is set with a new/differing email addy from what is already in a POPM, it would create a new POPM frame. otherwise, overwrite.

note: if multiple POPMs already existed, and overwrite were "on" the last POPM should be the one overwritten.

4.supplying a second toggle to only display ratings for the email specified in #2, (default: off)

note: when off, this would show any rating present by default, but would show what's specified in #2 as the preferred one for display. however, when "on" it would show only that one exclusively for display.

5. make the ML "refresh" its view based on how the above prefs are set/changed, when changes are made, on-the-fly.

--

that's pretty much it. the UI for this would be in the prefs, although a button or sub menu to an existing button in a skin to let you change the current email and/or other options discussed above would be cool, say via the top center file info window in bento.

Originally Posted by Fla_Panther

Mr. Sinatra's posts took the same path as my thoughts. Unfortunately DrO skipped a significant section as the overall post was tl;dnr. If I may, I'd like to bring back this one section:

interesting that the post was "too long" ...except of course for the part that interests you.

well, at least you aren't calling me names now.

[MrSinatra]

DrO, i really and truly sincerely appreciate it.

thanks!

this is great for me, and others who want to track who first rated a file. maybe one day it will lead to apps eventually supporting multiple POPM frames, so differing users can have different ratings in the same file, (based on their respective email addys).

again, thanks very much. i realize this is a "beta" feature and i won't complain if it doesn't work right, i'll just report back and let it be known.

[DrO]

sorry but tl;dr; i have to assume from the first few lines you're questioning what's going on.

as such it follows the same behaviour as 5.63, it just adds an extra level on reading / writing to use the custom value if changed from the 'rating@winamp.com' defaults.

so reading and writing will still use the existing values where possible, but if you have your custom one then that is the version that will be preferred which is the same as how it was doing when the plug-in was just using 'rating@winamp.com' for the Winamp version.

-daz

[MrSinatra]

ok, i await the release to test it out, exciting.

i wanted to post this because i find it so odd. i finally did WMP12 testing, and i found out some things:

afaict, no version of WMP ever supported half stars or fractions of a star, as set by users.

thats shocking to me, b/c i thought WMP did support half stars, in fact i thought i had done so myself in XP days on it, but i can find no evidence this is true.

WMP only supports fractional stars when displaying stars IN BLUE, which indicates that the ratings came from AMG or some internet based metadata source. if a user rates the file, the rating is gold/yellow in WMP. i do not know if blue ratings can be inserted into POPM or not, or if they are, if they use a different id string.

here are my WMP12 (win7) results:

222-255 = 5 stars when READ with WMP12, writes 255
160-221 = 4 stars when READ with WMP12, writes 196
096-159 = 3 stars when READ with WMP12, writes 128
032-095 = 2 stars when READ with WMP12, writes 64
001-031 = 1 stars when READ with WMP12, writes 1

the only difference between the wmp chart and the win exp chart, is the cutoff between 4 and 5 stars. win exp uses 223/224, while wmp uses 221/222. why? god only knows. (assuming i didn't make any mistakes in my testing, which isn't a given b/c the tools i have to use are crude) having said that, i would appreciate it if anyone were to doublecheck my work, and validate it or catch errors in it.

i post this as just FYI. i think winamp should continue to conform to the OS, not WMP, although the only small descrepancy between them is the 4/5 cutoff.

[ryerman]

Originally Posted by MrSinatra

.......the only difference between the wmp chart and the win exp chart, is the cutoff between 4 and 5 stars. win exp uses 223/224, while wmp uses 221/222. why? god only knows. (assuming i didn't make any mistakes in my testing, which isn't a given b/c the tools i have to use are crude) having said that, i would appreciate it if anyone were to doublecheck my work, and validate it or catch errors in it............

When using WMP 11 in Windows XP, I get the same results as you:

222-255 = 5 stars when READ with WMP11, writes 255
160-221 = 4 stars when READ with WMP11, writes 196
096-159 = 3 stars when READ with WMP11, writes 128
032-095 = 2 stars when READ with WMP11, writes 64
001-031 = 1 stars when READ with WMP11, writes 1

They do not match my results when using Windows 7 Explorer:

213-255 = 5 stars when READ with Explorer in Windows 7, writes 255
157-212 = 4 stars when READ with Explorer in Windows 7, writes 199
100-156 = 3 stars when READ with Explorer in Windows 7, writes 142
044-099 = 2 stars when READ with Explorer in Windows 7, writes 86
001-043 = 1 stars when READ with Explorer in Windows 7, writes 29

[Aminifu]

Hi ryerman & MrSinatra,

I guess its a mute point, since Winamp going forward will use what DrO listed in post #94 above, but I would like to know the exact methods you guys used to get such different granularity results with Windows Explorer in Win 7.

[ryerman]

Hi Aminifu
Here's what I did.
To discover the number of stars produced for each numerical rating:

1. Create 255 copies of a small MP3 file.
2. Using Mp3Tag:
- Select all files.
- Delete all ID3 tags.
- Write the Track Field using the Autonumbering Wizard.
- Write the Popularimeter Field using the "Format value" quick action, with %track% as the format string.
3. Observe the files in the Preview Pane of Windows Explorer with "#" and "Rating" as column headings. # (which is track number) is the same as the numerical rating in the Popularimeter Field.


To discover the numerical rating written for each of the 5 possible star ratings:

1. Create 5 copies of a small MP3 file.
2. Use the Details Pane in Windows Explorer to rate the first file one star, the second file two stars etc.
3. Observe the files using:
- Mp3Tag using a customized column in the files pane with the value %rating wmp% or
- a Hex Editor (converting the hexidecimal value to decimal) or
- some other application that decodes the POPM Field (I use Exiftool)

[MrSinatra]

Ryerman,

first, thx for your testing. however, i suspect something is wrong with your windows explorer results. they don't jive with my results, and they are just weird.

i have attached a zip file that anyone can use to reproduce my results. simply unzip, and then highlight each file in windows explorer to see the stars. the filename and track number represent the raw POPM numerical value. (you can confirm that using any ver of mp3tag and the %popularimeter% column)

notice i included two "wmp" files so people could also test WMP, and validate the odd difference in the 4/5 stars cutoff from windows explorer.

just fyi, i used win7 home premium 64 bit to do my testing.

[MrSinatra]

wmp12 view of my attachment, showing the weird cutoff.

[ryerman]

Originally Posted by MrSinatra

......i suspect something is wrong with your windows explorer results. they don't jive with my results, and they are just weird........

Maybe

Here is a screenshot of your files shown in Windows Explorer on my computer.
I have highlighted 3 files that do not agree with the chart in Post #112.

If a different number of stars are shown on your system then there is something happening I can't explain.

I am using Windows 7 Home Premium with the WMP feature turned off.