|
Post by Earl Purple on Feb 28, 2014 1:00:39 GMT 1
So there are 2 songs left.
*2. Breathe In, Breathe Out - Tich
This song spent one week in the chart at #3 at the end of 2012 and if you add that in, it moves to #1, in fact it moves to #1 of the decade. Even then it is only 4 points above the song that is #1 on that chart, and 2 ahead of "Dave" by Missing Andy which was the #1 of the decade (but only the #3 of 2012 also because of a couple of weeks in 2011).
This is the song that really introduced me to Tich. It was one of a few songs she played when I went to see the Jubilee concert in Hyde Park, primarily because the Jive Aces were headlining it (they went on stage several times compared to most of the other acts who only went on stage once). The Loveable Rogues also appeared on stage in that show.
On 29 November 2012 I was one of those invited to attend the launch of the Candlelight EP and that is where the photo was taken that is my profile avatar here.
On 13 February this year I went to see her live at The O2 Academy in Islington, where she performed her "new uptempo" version of this song. Not sure I agree to her changing a song that is perfect the way it is, although the string arrangement here was newer than the all-piano arrangement I heard the first time and appears on another youtube video. Nick Jonas co-wrote it with her.
|
|
TheThorne
Member
*Hillside, slip and slide, feel the pain, it's no surprise!*
Posts: 27,529
|
Post by TheThorne on Feb 28, 2014 7:45:51 GMT 1
She hasn't released her album yet has she? I think she is alright but if it wasn't got all the attention you have given I probably wouldn't have given her a second listen .
She is one if the rare acts that have benefitted from advance listening in your chart. , give a song a week to get your attention from a Spotify playlist or a song that you have heard for ages before it comes out and they will always win , example in my chart would be at the moment the new Foster The People which I heard first 2 months ago and it's still not out and it will go straight into my top. 5 when it does .
But back to Tich she is alright can't say much more and again I try to think if this song and loads of other breath in breath out songs come into my head like Luce Silvas or Delta Goodrem not her song hmmm
|
|
|
Post by Earl Purple on Feb 28, 2014 17:00:32 GMT 1
Yes she definitely gets the advantage of my advance attention on the song. This song I knew for a while before, Dumb I saw her perform live and also saw her promotional video. At the end of the 3-weeks where I don't listen to music, the first thing I did was watch the video of Obsession. Although the only one to enter my chart at #1 was the Christmas one but that probably was partly due to the short chart-life of Christmas songs that I will enter them higher.
I have spent much of this week finding ways to search and filter more effectively but I now need a magic program that can actually listen to the songs and decide which ones are good and bad. Unfortunately, unlike using an API to search for metadata to see if a song is really new, if it's a single and what genre it is which is not beyond my capability, getting a program to be able to determine what I am going to like is likely to be near impossible.
Number one coming up in the next post in a minute or two. Firstly some stats about the top 10 or so:
The totals were:
330.9577922 359.2545394 369.880114 398.8516746 406.6331753 427.4320157 449.6442781 450.9496841 459.5781475 498.2567374
Tich's week at #3 in 2012 moves that song up to 502.4352904. I sometimes consider a song crossing 500 points as "going platinum" but it isn't really a fair comparison.
Tich has the highest placed 2nd song with Obsession. She doesn't have the highest placed non-#1 as that is Butterflies by Sensible Sam & Japan.
|
|
|
Post by Earl Purple on Feb 28, 2014 17:12:47 GMT 1
1. Hard To Believe - Records And Tapes
This is a duo I know little about. They are from Los Angeles as their facebook page gives this info:
I have never heard of Dreaming In Stereo.
This is the sort of song I hear and really like, have to look up a few times who it is on it, and then see they released something prior to this I hadn't heard of or charted an start thinking that if I couldn't see that they had something out before, what hope is there of people being able to find the music they are looking for.
I am of course resigned to the fact that there is so much music out there to discover, far more than one is capable of listening to and thus will miss music, and whilst I can compile lists more easily, and match artists against those who have reached my chart before (which I am doing now which is why I am getting far more returning artists to my chart) I do need some help in discovering.
This song was in a two-way battle with The Damn Jammage which entered the same week. This one entered higher but got to #1 second, a bit like the UK chart battle between Madness and Adam Ant in 1982. This song also subsequently kept Sunrise Avenue out of #1.
It's good to see people "check out" what I am charting high, which is perhaps the primary reason to make a chart, and I already do look at some other people's charts to help me build my own playlists. I should perhaps do that even more, perhaps with some automated search.
I am hoping now that the 2014 countdown will yield as many good songs as this one has.
|
|
TheThorne
Member
*Hillside, slip and slide, feel the pain, it's no surprise!*
Posts: 27,529
|
Post by TheThorne on Feb 28, 2014 17:46:57 GMT 1
Guess you need to relate some heuristics to the genres you usually like but yeh that is some pretty advanced AI requirements can't really apply logic to taste but I think the list will be manageable especially If you exclude various artists as the new version of pantsentient league Spotify releases has got bogged down with too many compilations
|
|
TheThorne
Member
*Hillside, slip and slide, feel the pain, it's no surprise!*
Posts: 27,529
|
Post by TheThorne on Feb 28, 2014 17:53:08 GMT 1
As for your number one I have heard of both bands but I don't think I have charted Records And Tapes and will need to check my downloads to see if I have any Dreaming In Stereo but I think they are both bands that I have seen in new band lists in Nme but never looked further , will listen to your number one later just to check but I think I missed it completely
|
|
|
Post by Earl Purple on Mar 2, 2014 0:00:50 GMT 1
On Spotify they list one self-titled album by Dreaming In Stereo from 2009 and no singles. Assuming there were no singles it is a likely reason why I didn't get anything off them.
I did a "run" on Friday morning of "new" which eliminates automatically anything that I found on the previous Tuesday morning run. I then removed genres I didn't want and all the ones with a foreign title.
The matching with iTunes already lets me eliminate older ones.
Even with all this, those few days yield a list of 447. The genres are not really properly descriptive, I have "alternative", and "pop" and "rock" and "indie rock" and "singer/songwriter" as the main ones. Some others like "Country" and "Jazz" I am unlikely to want but interesting to see what is in there.
For example, the song that is probably my "Single of the week" this week came up as "rock" in the genres and I wouldn't describe it as rock, I would call it alternative/pop. Of course iTunes probably can afford to pay staff to listen to the songs and put a more meaningful description to them. I can't.
|
|
TheThorne
Member
*Hillside, slip and slide, feel the pain, it's no surprise!*
Posts: 27,529
|
Post by TheThorne on Mar 2, 2014 0:12:37 GMT 1
The metadata is added when the digital track is made , you should be able to extract it as Pansentient does , iTunes doesn't actually listen to all the tracks and 447 is manageable I am more concerned about missing something by somebody I like or a song that I know is coming out soon than having to listen to every song that comes up, and maybe you could do it daily that would be even better for me
|
|
TheThorne
Member
*Hillside, slip and slide, feel the pain, it's no surprise!*
Posts: 27,529
|
Post by TheThorne on Mar 2, 2014 0:28:45 GMT 1
Oh and Dreaming in Stereo have had. 2 albums and at least 3 singles called. 'Home', 'steal this song' and 'fill my sky' whether they were released in the uk or were. 'Proper' singles ie ones that get a separate release rather than just been promoted and cherry picked of the album I don't know.
|
|
|
Post by Earl Purple on Mar 2, 2014 0:29:32 GMT 1
Pansentient extracts the genres from last.fm and not from iTunes like I do.
I have a list of every artist that has reached my chart in the last 5 years. I could make it go back further. I match my long list against that to find any that match. Those get priority on my playlist.
447 was a list just from Tuesday morning to Thursday night (not Friday morning). If you want the list I will have to e-mail it to you for now. I'll have to see how much interest there actually is to generate these, possibly off the spotify and iTunes community areas.
I could send you the lists as I generate them, totally unfiltered (other than karaoke etc.). At the moment I am also filtering out anything that has the word "mix" in the title, and anything where the lead track is longer than 6 minutes. I can of course filter a lot less, produce lists with albums too, etc.
|
|
|
Post by Earl Purple on Mar 2, 2014 3:02:18 GMT 1
"Big When I Was Little" was the only NM #1 of 2013 that was a UK top 40 hit.
Dumb finished higher though in the chart of the year, at #15. Alex Day would have been one place higher than it though counting the last week of 2012, the week in which he also was in the UK top 40.
Big When I Was Little peaked at #12 in the UK, Dumb at #23 and Stupid Stupid at #25.
|
|
TheThorne
Member
*Hillside, slip and slide, feel the pain, it's no surprise!*
Posts: 27,529
|
Post by TheThorne on Mar 2, 2014 8:11:07 GMT 1
Yeh cutting out mix is a good idea but the new Horrors single is over 7 mins so does that mean you don't allow it. That's a shame as it is awesome
Your program sounds brilliant the comparing it against artists thing is that complicated as well , what you would eventually want I'd it too compare it to every artist you have ever charted it or least everyone in your Spotify /iTunes account as I have 40000 songs in my iTunes account and several thousand in Spotify that could be a basis for the AI routine , you can tell I am studying AI right now it's the last part of my ICT degree hehe
I will wait till you have a web based alpha I would love to play about with it once it is there
|
|
|
Post by Earl Purple on Mar 2, 2014 11:39:36 GMT 1
At present I have no web-based version, it is all scripts on my ubuntu box. I don't even have a proper database at present although I plan to possibly put in a MySQL or possibly even SQLite backend.
At the moment it is there for my own benefit only, but then I only wrote it last week. The spotify API is quite complex and the only way to search for new releases is tag:new search. You can actually type that in as a search in the GUI itself but it will return a lot slower as it downloads all the pictures for you and then takes ages to scroll down.
The difficulty with the API is that, although it only accepts single-threaded requests, you are expected to wait on events, and handle them as they come in whilst waiting for your results to return.
Of course I could put the iTunes matching onto a different thread in the same process, which would thus finish everything quicker, but I decided to do them one after the other for now.
I am assuming that the "tag:new" results stay fresh for no more than 1 month maximum so I don't have to check against every album uri ever, just those from the last month. These are loaded into memory at the start and hashed, so the search is relatively fast.
The iTunes search is relatively simple by comparison, a straightforward HTTP_GET and parsing the JSON response, and there are libraries to enable you do that. My own process sticks with C++ and is using the POCO library.
Many music sites have APIs. The artist comparison against my own chart is done later when I have the full list. As for the filter, that is again only for my own chart. If I'm going to develop this for the use of others I can try to allow customised filters. The problem is though that I only want to run the process once and cache the results (in MySQL or whatever) so users would be doing queries against my database which means I would have to store all the results without doing any filtering at this stage.
The question then remains whether or not I do the processing local then upload to the cloud-based MySQL database used by the web-based front-end.
|
|
|
Post by Deleted on Oct 2, 2017 11:04:55 GMT 1
|
|
|
Post by Earl Purple on Oct 2, 2017 11:12:50 GMT 1
The chart of the year is always based on points and Tich is #1 of the year if you count its week in 2012. The male singer in Records & Tapes is Fernando Perdomo who had a couple of solo hits the following year. Otherwise they remained one-hit wonders.
"Hard To Believe" came up on "random oldies" recently.
The biggest two singles of the decade so far were both hits in 2016. The artist who was #2 of the year in 2016 was #1 of the year in 2011 but their 2016 one was a far bigger hit.
|
|
|
Post by Earl Purple on Oct 2, 2017 11:19:58 GMT 1
She's now a songwriter writing for other artists and often collaborates with Jez Ashurst, who wrote one of my biggest hits of 2011 ("Scarborough") and also performed on it as Farrah was a band he was a member of. "Secret Love Song" was originall co-written by those two and later had new parts added to it. Still I do not see that as a reason to remove her music from her past. Last year I decided to do a "rate" of all my most recent Christmas number ones and a few lesser ones. "Love On Christmas Eve" came in 1st place followed by the Dropkick Murphys one of the previous year. 2011's "Christmas With Bing" was ranked the lowest of them with "Crazy Christmas" by Kieron Goss (2008) next although both of those are way better than some of the Christmas hits of 2009 (other than the actual #1). I've had a Christmas song at #1 on or close to Christmas every year since 2007 except for 2012.
|
|
|
Post by Earl Purple on Oct 2, 2017 17:32:23 GMT 1
Some of the discussion in the topic here is based on March 2014 which is when I wrote a lot of the programs I still use now to find new music, although the Spotify API changed enormously and is actually simpler
Last week it seemed broken and returned loads and loads of hits from way back as well as new music. So as Spotify also gives a release date I may need to filter on that too, and have a minimum date on their list or cut.
I have a program called "cleanSearchResults" which removes many of them.
iTunes limits me to 20 a minute which is why I need fewer to get through to that stage.
All those are written in C++, the youtube search one is written in Python. It would be possible to write all of them in Python. That one wasn't written until 2015, so I have no idea how I managed before I had the video restriction, or maybe just there are far more musicians now throwing their music onto streaming sites since it took off big-time in 2014. I think a lot of them were weary of Spotify at first and now they know they are not bigger than the system and if their music isn't there they simply won't get people listening to it, let alone liking it enough to want to buy it or see them live or whatever.
Spotify metadata also has genres but not all music has it, and it's too wide to make it easy to filter on.
|
|
|
Post by Deleted on Oct 2, 2017 17:37:07 GMT 1
I always have a dream about such tags for searching as "female singer", "girlband", "the group with female vocalist" (like Texas or Blondie), "mixed group" on Discogs, Rate Your Music and other music databases sites.
|
|