h a l f b a k e r y
A few slices short of a loaf.
add, search, annotate, link, view, overview, recent, by name, random
news, help, about, links, report a problem
or get an account
Analyses voice in the structure of a song, compares with lyrics parsed from 'the internets' and automagically displays the words to the song in the style of the Sign o' the Times musicvideo (see YouTube link).
<added later> If distinguishing voice from music might be difficult for certain kinds of
music, a prepared, manually fine tuned, timecode from another user (in some online database) is the first choice for the application.
Just like you keep .srt files next to .divx movies, you then store those .srt files with timecodes next to your mp3 collection.</addition>
(?) Sign o' the Times
the video on YouTube [rrr, Nov 28 2006]
lyrics database example
just one of the many databases with lyrics from popular songs that listen to the query lyrics in combination with the title of the song [rrr, Nov 28 2006, last modified Nov 29 2006]
The Kleptones mix music and lyrics from different sources
The Kleptones pick a song and filter the music out and mix it with another song with the vocals filtered out. Listen for example to how they combine The Clash with Rock the Casbah and David Bowie in the first song on their album '24 hours', which you can download for free under a Creative Commons license. [rrr, Nov 28 2006, last modified Nov 29 2006]
(?) tascam vocal trainer
vocal "eliminator" [bleh, Nov 28 2006]
get subtitles for movies
There are many websites which offer text files containing the subtitles to popular movies. The timing is contained in the text file in some standardised format. Just place the file with the subtitles in the same directory as the divx and play (with the VLC mediaplayer for example) [rrr, Nov 28 2006]
(?) Identity 2.0 presentation
Presentation in Lawrence Lessig style with practically every other word illustrated with a picture [rrr, Nov 28 2006]
this app does a reasonable job of, ahem, looking up the lyrics to the current song in a database. [neilp, Nov 28 2006]
(?) Lyrics widget for OS X Dashboard
It's been a few years since I last tried a widget in Dashboard to get the lyrics for the iTunes song I am listnening to, but I now found one that both looks good (it includes the cover art) and does a good job. [rrr, Nov 29 2006]
displays lyrics in combination with cover artwork when the iTunes Visualiser is switched on [rrr, Dec 09 2006]
||thats asking a lot from a music player,
dont ya think?
|| is it supposed to store the lyrics to all
your songs on your harddrive or stream
then from some mysterious source on
||also, how is iTunes to analyze the voice
in a song? its dealing with a simple
stereo wav or mp3 file, not individual
tracks. that essentally makes separating
the voice from the rest of the mix
damned near impossible unless the
voice is panned dead center and
everything else is 100% left or right.
||bone withheld hoping for further
||Actually, I do not expect iTunes to do it all. Just some extra application that is tied to iTunes. If iChat can read what song I am playing, so can any other app I guess. Also see the links to the left of this as an attempt to answer your question and to avoid the smelly fish.
||nice links, but again, is this extra
application supposed to do all this real
time or do it when you import and save
like 3 files per song, then incorporate
them all when the vizualizer is turned
||also, people can spearate lyrics from
music by carefully aplying filters and eq
to it dynamically , but a computer
would have a
helluva time trying to follow that with
an algorythym (sp?). people can do it
mainly by ear.
||I'd love to see the technology, right now
it doesnt exist. I work in a music store
and we sell product that is suposed to
"eliminate vocals" (linky) but all it does
is kill then center channel, where many
vocals are mixed. often there are
additional vocals (harmonies, reverbs,
etc) mixed to one side or the other. they
make similar pieces to elimnate guitar
or bass, but those are just EQ tricks
and are equally dissapointng.
||Good suggestion, let the app do the analysis of the tracks and retrieving of lyrics at some other time. The end user should also be offered some editing screen in which you can fine tune the results. Other fonts, play with sizes, colors and so on.
||For tracing the vocals it might be helpful that the lyrics are known because the title and artist are in plain text. The bpm and the length of the song are also known. All that together should be enough to get the lyrics more or less timed correctly to the vocals. And the end user can do some fine tuning and post the result as an .srt file back on the web, to share with other users. Let a new subculture of user generated content evolve out of this, in the same style of that rage of 'demoing' (does it still exist?).
||No, this is all too simple. The idea needs to be more complex.
||The application should review the lyrics and construct a video from stills and video from the net.
||ok, if the end user imputs it using
timecode as he listens to the song, that
is a lot more feasable, as it doesnt
require an (currently) impossibly
complex computer program. I dont
think even knowing the length of a song
would help a program as it doesnt
account for solos, long interludes, and
||so what would probablly work best
would be a program that plays the song
and allows you to type (or drag the
downloaded lyrics) along timecode
while listening to the song. that way the
human ear can interprit the dynamics of
the music and make everything line up
better. maybe hit the spacebar to inser
the next word at that point in the
timecode, hold the bar down for
sustain. something liek that. other
expressiveness (color, font, boldness)
could be controlled with other keys.
||bun if you take out the computer
analyzing the vocal track.
||And translate into Russian and then into
Japanese and then back into English.
||> a video from stills and video from the net <
|| Perhaps not that ambitious right away, but just like in the Prince video certain expressions are replaced with icons, the user should be allowed to insert emoticons and or references to a directory of JPEGs that replace certain words. Just like that great Keynote presentation on BoingBoing a while ago. A kind of Lawrence Lessig on steroids presentation, practically every word was illustrated with a picture. It was on Identity 2.0, see link.
||I very much liked that Prince video. It's
interesting that glyphs arranged in
groups can have such embedded
meaning, and yet also look interesting
and attractive, too. Mostly. Who'd have
thought that those 'letters' that have
names that don't sound like their shape,
and that don't necessarily correspond
very closely with the sound we're
supposed to make when we see the
glyph, could be combined in such
manner as to make groups that have
even higher meaning. Then combine
those groups of glyphs into even longer
groups of groups, and you can express
relatively sophisticated abstract
notions. Amazing. Then move them
around a bit and make them different
sizes and colours, and change these
parameters in time to music, and you
add almost no further meaning at all.
But that's okay, there's a little chap
singing the same words, so you get the
same information through two distinct
||Nice review of the Prince clip! Makes you watch again. Closely.
||What you want is a karaoke file, or karaoke MIDI file, with just the lyrics extracted, and parsed by your iTunes plug-in in sync with the track.
||BunsenHoneydew said it right. It should
not be hard at all to incorporate this data
into any track. It's just red tape that's it.
||It is a little baked in the meanwhile. Some German wrote 'Coverversion'. See link.