One of the announced features coming up in 1.1 is the player's ability to weight the shuffle so that songs you already heard recently end up near the end of the list. I'm looking forward to this feature.

It was also mentioned that the player has already been keeping track of the recent-ness of a given track. But it was also said that this information had to get thrown out recently in favor of a different format.

My question is: Was the new format implemented in 1.01, allowing the feature to begin working instantly when we get 1.1? Or did the format change happen after the release of 1.01?

___________
Tony Fabris

The format change was in 1.1 I'm afraid; it's even changed since versions I was using early on in 1.1.

It'll be stable by the time 1.1 ships, and your stats collection will start from there.

Hugo

Tony: ...weight the shuffle so that songs you already heard recently end up near the end of the list.

Hugo: It'll be stable by the time 1.1 ships, and your stats collection will start from there.

Does it collect stats regardless of the circumstances of play? In other words, if I turn shuffle off, and then burrow down through the playlists to find my favorite album and play that album start to finish, does that reduce the likelihood of tracks from that album showing up near the start of a randomized (i.e., shuffle on) playing of the entire playlist? Or does it only collect statistics to weight the shuffle while playing in shuffle mode?

Does it keep statistics on skipped tracks? If so, what does it do with that information? Does it move the skipped track higher up in the shuffle next time because I haven't played it recently? Or does it move it lower in the shuffle because apparently I don't like that track?

tanstaafl.

"There Ain't No Such Thing As A Free Lunch"

Peter's probably the best person to answer this, but yes - it collects statistics for every song played, shuffle mode or otherwise. There are multiple shuffle types in 1.1, the default ISTR is least recently played - so yes, if you play an album directly, it's likely to end up near the end of an all-player (or even all-tracks-by-that-artist) playlist.

It keep stats on skipped tunes, though I don't think anything is done with that yet.

Hugo

The stats are accessible as fields in Emplode 1.1 searches - so you can search for, say, tracks that you have skipped past more than 10 times and then review and delete them. I think that's the main use for that particular field, although it may be cute to be able to use it in a Wendy filter (lose all tracks that I regularly skip).

Rob

This begs the question: How does the player decide the difference between a "skipped" track and a track that you listened to a significant portion of before moving on?

___________
Tony Fabris

I think it'll take a look for how long the song was played and if that is more than xx seconds then the song is not counted as skipped. The question is how many seconds.

TeeMcBee
Got my Mk2! # 080000143

Currently, xx = 5. This may change.


Roger - not necessarily speaking for empeg

Currently, xx = 5. This may change.

There are several counters per track, including last time played, number of times played, and number of times skipped. If you listen to a track for less than five seconds before skipping on, it won't count as played. However, it will count as skipped.

As someone's already said, the default random shuffle mode in 1.1 will be based on time last played (in Mark 2s) or on number of times played (in Mark 1s, which don't have a real-time clock).

The 'skipped' information isn't used in the default random shuffle. The player supports custom shuffle modes, but I'm not sure that defining custom shuffles in emplode will make it into 1.1.

Peter

It would be nice if it kept a average percent of song listened to statistic. That way, each song will have a percentage. If you skip it immediately regularly, it will register a 0%. If you never listened, it would be null%. If you listen to half of it before skipping then 50%. That way, you can easily figure out likability and unlikability of different songs. This should be in addition to tracking the number of times listened to, because that could be used for weighting purposes.

Calvin

As someone's already said, the default random shuffle mode in 1.1 will be based on time last played (in Mark 2s) or on number of times played (in Mark 1s, which don't have a real-time clock).

Uh oh, I don't get the "recent" feature?

For me, I think the number-of-times-played would only be a pale substitute for the "recent" feature, not behaving quite the same way...

Sigh.

___________
Tony Fabris

Uh oh, I don't get the "recent" feature?

Well, Tony, here's your excuse to put that old, obsolete, technologically lacking antique of a Mark I into your wife's car and join us in the 21st Century with your very own Mark II. Don't let time pass you by! Oh, that's right, you wouldn't know about that because your empeg doesn't have a way to keep time, does it?

I have found that the empeg has changed not only what music I listen to, but how I listen to it. Before actually using the empeg, I didn't think that the shuffle feature was anything I would be particularly interested in. Tony's post here opened my eyes a little bit, but I didn't fully believe what he was saying until I had the use of my own empeg. When the modified shuffle feature (i.e., less heard = higher priority) becomes effective, it will make what Tony said even more true. Think of the empeg as your own personal radio station that only plays music that you really like (after all, if you didn't like it, you wouldn't have put it in!) and that at the same time makes sure you get to hear all of your music, not just the music you can think of off the top of your head.

This modified shuffle will be one of the most important and useful features of the empeg. [he said, pontificating...]

tanstaafl.



tanstaafl.

"There Ain't No Such Thing As A Free Lunch"

There are several counters per track, including last time played, number of times played, and number of times skipped.

Will there be a way for the user to reset those counters should he for whatever reason decide he wanted to start out with a clean slate? Say, for instance, if somebody bought a used empeg? Or if he got tired of always hearing songs he hadn't heard in a long time and wanted to hear some of the songs he liked that were always low in the priority list because he played them a lot?

tanstaafl.



"There Ain't No Such Thing As A Free Lunch"

Will there be a way for the user to reset those counters should he for whatever reason decide he wanted to start out with a clean slate?

I hadn't thought about it -- is that really useful?

Say, for instance, if somebody bought a used empeg?

Delete the other guy's songs and upload some of their own?

Or if he got tired of always hearing songs he hadn't heard in a long time and wanted to hear some of the songs he liked that were always low in the priority list because he played them a lot?

Completely random shuffle isn't going away. Biased shuffle will probably be the default, but you'll be able to choose (from the player menus) completely random shuffle as an alternative.

If you really wanted to wipe the slate clean you could delete the songs and re-upload them?

Peter

I hadn't thought about it -- is that really useful?

Don't you just love it? You invent a whiz-bang new feature, and then your users come up with other necessary features as a result.

For what it's worth, I think that having the ability to select a non-biased shuffle solves all those problems.

___________
Tony Fabris

A used empeg should be completely clean, since in theory you're not supposed to distribute your music...

Calvin

If you really wanted to wipe the slate clean you could delete the songs and re-upload them?

Is this your real opinion? I hope not. Cause uploading 6GB and more (much more) would take a whole lot of time (6GB are well over 2 hours calculating high 750kByte/s over IP). Updating the database with about 600-800 songs wouldn´t be especially fast, but _way_ faster than uploading them all from scratch.

Anyway, the biased shuffle really asks for another feature: user profiles (make it possible to log in as a specific user that has different counter values than another user). And different counter values for in car / indoor use, because I have quite different listening habits for those two situations.

cu,
sven
(MkII 12GB blue, #080000113)

Updating the database with about 600-800 songs wouldn´t be especially fast, but _way_ faster than uploading them all from scratch.

I guess we could move the files out of the FIDS directory, rebuild the database (by starting the player), and then move them back and rebuild again. Would that reset the tunes' values?


_~= Dearing =~_
"WAY too happy about having #99."