Just by way of sharing concrete tactics (rather than, I guess, the abstract strategy I usually share (?)), I thought I’d write about how I learn songs using the SRS.
Keep in mind that this is just what I do right now. Yes, I am the Great Khatzumoto, but you know what? Really I’m just a 27-year-old boy who drinks peppermint tea and plays with his cats. I don’t know jack about jack. This is just what was most fun and least annoying for me. You’re bound to have a better idea and I’d love to hear about it if you’d like to share 😉 .
Put another way: a lot of AJATT strategic principles are universal, I think. But the tactical stuff is totally a matter of “what works for you”. I mean, we might not even be running the same OS, so…you know. Anyway, here we go!
The Steps
- Have a song that you love, love, love and wish you could sing along to
- If at any point in this process you get bored…stop. The worst thing you could do for your Chinese/Japanese/any language is start to associate it with boredom. That there is the gateway to failure. Having fun with and in the language is the name of the game.
- Get an mp3 file of the song.
- Split the file into 10~30-second clips with ~5 seconds of backward overlap
- I add a 5-second backward overlap because a split on strict time boundaries is bound to be imperfect in that it’ll cut right in the middle of something good.
- Adding the overlap provides a way to automatically compensate for this without going through the psychological and computational heck of attempting to split on something like silence-points.
- I use EZSoftMagic’s MP3 Splitter & Joiner for this automated splitting
- They’re not paying me for this endorsement, but they should 🙂
- If you know of any other software that does a good job at this, feel free to share in comments.
- The reason we split the file and not just throw the whole thing into the SRS is because we are trying to do what the SRS does best – optimize the management and memorization of discrete chunks of information. Throwing the entire song in there is (1) boring and (2) defeats the purpose of even having an SRS.
- I add a 5-second backward overlap because a split on strict time boundaries is bound to be imperfect in that it’ll cut right in the middle of something good.
- Get the lyrics of the song
- Put the audio clip into the SRS on the back of the card
- Put one line or less of the lyrics on the front of the card
- i.e. the lyrics of a segment of the 10~30-second clip, not of the whole clip
- Put the lyrics of the whole 10~30-second clip, or of the entire song, on the back of the card.
- I prefer putting the lyrics of just the whole clip because it’s easier to read
- But sticking the lyrics of the entire song on the back could save you a lot of fiddling
- Do your reps.
- The task is to read aloud or sing the line/segment of the line of the song
- Check your “answer” against the actual song clip
- Final note: if any of this feels like too much work, then stop. Abort. Delete. Whatever. Because you obviously don’t like the song enough. You may like the song, just not enough, not that much. And that’s fine. Remember, the idea is to be like Soviet Russia: let the media motivate you — that’s its job. All you have to do is put yourself in the path of the media.
Sample Card
FRONT
男兒當自強
[Youtube]
BACK
[media: naam yi dong ji keung- 007.mp3]
廣闊浩氣揚 既是男兒當自強 昂步挺胸
jìshì【既是】
…であるからには．…である以上．
gei si naam yi dong ji keung
Benefits of this method
- Over time, with very little effort, you learn the entire song
- As per SRS principles, the parts of the song that give you the most trouble – and that therefore need the most practice – will get seen the most
- Ever notice how almost everyone knows the chorus of a song no matter how complex the vocabulary? (I remember being about 6 years old and singing Bobby Brown’s “it’s my prerogative!”). That’s because the chorus gets repeated so much. SRSing the song turns the entire song into a “chorus”, in that all the parts of the song will get repeated to the degree necessary to ensure their memorization.
- No need to fiddle with carrying lyric sheets in your bag or on your computer – it’s not like you can ever get them out on time anyway.
- Even after the song stops getting playtime on your mp3 player, the SRS will ensure that you keep getting practice with it. This is a microcosm of how the SRS is a powerful partner to an immersion environment – even after you stop immersing in, say, technical documents from a certain field, the SRS will guarantee you keep getting the practice in that field that you need to retain your proficiency in it.
Funnily enough, I’ve been considering entering guitar tablature into my SRS as a way to get around how boring I’m finding it at the moment to learn to sing new songs in new languages. The method I came up with may also be applicable to learning song lyrics. It tests pretty much rote memorisation of a thing without concern for any understanding of its meaning.
FRONT:
1. Either an audio clip from the song, or
2. A line from the song written down, or both.
BACK:
The NEXT audio clip/line from the song.
The task is to hear/read one line, and be able to produce (sing/say/write-if-you-really-want-to) the next line.
I’ve been thinking about adding sound clips to my song lyric cards for a while, but have never really come up with a good way of doing it. Always thought it would just be a lot of faff to chop up the mp3 etc.
I just take the lyrics from a song, put them into notepad and make it into a anki import file, very very quick to do, but having the song clips would make a world of difference.
Maybe there is a way to use the subs2srs software to make it quicker? Turn the lyrics into a .srt file???
Good stuff. I use something similar for learning poems that I want to have ready access to. Basically:
FRONT:
Blah blah blah blah
[Missing line]
Blah blah blah
BACK:
[The text of the missing line]
Works like a charm. And then if there are any intra-line things that I mess up a lot I’ll make cloze deletion cards for them. It’s remarkable how easy it is to remember structured things like poetry and song lyrics.
On a tangental note, WTF is up with so many Cantonese singers singing in Cantonese-pronounced 国语? It’s sort of infuriating. Hip hop is pretty much the only place I’ve found any real colloquial Cantonese lyrics, and even there there’s not all that much.
I did this when I first started AJATT a year ago. It was a little frustrating at the time because one line of rap can have a lot of vocab that you don’t know. Plus, again if its rap, the line might not have much meaning at all which can be a little frustrating too. But… I think I might try it again… maybe. 最近見つけたスチャダラパーがメッチャ好きだから彼らの曲を使えばいけるかもね。
a great online mp3 splitter.
mp3.cut.net
Technical quesiton: how would one apply this idea to youtube videos or other audio that’s streamed online, that you don’t have a mp3/wav of saved on your comp?
I tried recording what I needed with audacity, but there seems to be some kind of flaw with the mic/recording on my comp that won’t let me do it. I also used to use Keepvid.com to download youtube vids (and would then convert them into mp3 format seperately & add them to my iPod), but keepvid has decided that they’re not going to let people use their services for free anymore, and I don’t want to give my credit card # to strangers on the internet. Other websites similar to keepvid either don’t work or try to give me viruses.
Anyone know anything that might help? Simple explanations would be appreciated, as I’m not technologically-inclined.
In regards to text-only SRSing of songs to learn the vocabulary contained in them (and not necessarily the actual song itself): I have found that the normal SRS approach of sentence on one side/kanji readings and translation on the other does not work, the reason being that you just sort of “hear” the song in your head as soon as you start reading the card. This phenomenon makes learning kanji readings from song lyrics nigh impossible with this card format, as you will be unable to read the kanji outside of the context of the song.
My solution is to give the kana on the question side and write out the kanji readings. I have found that this works quite well.
Oh wow… that turned out quite TL;DR… anyway, Khatzumoto I have been reading your site for a several months now, but this is the first time I’ve seen fit to post anything. You rescued me from a textbook-induced slump that almost made me quit Japanese. I’ve completed Heisig 1 and have been doing sentences/mild immersion for a couple months. Your method most definitely works. You are a terrific motivational writer and a stunningly good looking human being. I believe that thanks are necessary for the massively positive effect your blog has had on my learning, and in some respects my entire life.
So: Thank you, Khatzumoto.
Wow, more TL;DR. anyway, you’re f***ing awesome. keep up the good work.
I think this is a great idea. alternatively downloading the mp3 version of anime clips on youtube is another great way to do this. also check out my channel 🙂 I’ve japanese subtitled some anime just for you..yes im just that generous. srs it up
www.youtube.com/user/FranksAJATT?feature=mhw4
Thanks for sharing this! I love learning languages through music, so I’ll have to give this a try 🙂
I highly recommend Audacity for splitting up the MP3s. It’s completely free and very straightforward.
Although it’s no help for Chinese and Japanese, for those learning European languages lyricstraining.com tests you on your knowledge of the lyrics for popular songs in those languages, synchronised with the music video.
Nice idea!
I will try it today and looking forward to good results. Maybe this is an good strategy to overcome the problem i have with lyrics. Being too picky about the mining thing. I just get 2 sentences sometimes from any lyrics. However, what do you do if you don’t understand some lines or parts? For example i really love zeebra’s street dreams, though there are some parts i really have no clue how to understand them. (ぜってぇ誰も真似できねぇ俺のライフ
掴めNo.1ヒップホップドリーム 胸はれ誇り高き日本人) Shall i just look up the vocabulary and enter them, or shall i look up the vocabulary, the grammar and then enter them or finally should i just skip them? Oh my gosh xD Zeebra is so damn cool. I hope to get more from lyrics and to increase my daily sentences picking thing^^
As always thank you very much Khatz
P.s.: Have anyone problems with entering pictures and audios into SURUSU like me? Would be great to just use one srs <.<
Greetings,
@Maya: There are sites on the internet that will convert the audio from youtube videos to MP3. I’ve used www.video2mp3.net/ myself. You take the URL from a youtube video (for example: www.youtube.com/watch?v=bVDZH6Gtg08 ) and enter it in. It will then convert it and let you save an MP3. There are other ways to do this, you can google “youtube to mp3” for it if this site doesn’t work well for you.
Then you can use the other methods listed in this article.
I hope this is helpful.
Another great program for getting audio clips as well as splicing is Audacity. I’ve used it for a while, audacity.sourceforge.net/ is the link to the thing. There are tons of youtube vids on how to use it too, so have fun.
As for tracking music through the SRS, This is what I did before sentences and even rtk (though albeit after kana so I could at least figure a lot of it out). I love music and this was a way I could get a lot of immersion. I love to sing (yes I’m one of those people who sing in the car, shower, cooking, cleaning, ect.), and I love to sing Japanese music even when I have no clue what I’m singing.
My daughter loves me to sing Japanese to her too, which gives me incentive to do it right, since I have audiences. Good luck everyone!
If you want to download YouTube videos then I would recommend the addon DownloadHelper: addons.mozilla.org/firefox/addon/3006
It also works for a hell of a lot of other streaming sites too, you can save it in either flv or mp4, it’s very simple to use.
Then just convert the file to mp3. To do this I just use VLC player (www.videolan.org/vlc) since that’s already on my computer and it’s free and open source, saves me getting something else.
If you want to know how to do it, just go to File Convert/Save or Media -> Convert/Save then choose file, then from profile choose MP3, tick File and choose save location and press Save.
You can also mess around with some of the options if you want to and save it to 192kb/s, 320kb/s or whatever else you wanna do to it.
Or you can just use another program instead.
Interesting idea, but way too much work for me (editing/splitting audio? that sounds like my real job) 🙂
I usually just listen to songs I like… a LOT. Especially if I fall in love with a certain artist, I usually end up liking all their tunes and play them over and over.
I usually pick up about um honestly, 10-20% of the lyrics in certain phrases (not necessarily the chorus) which stand out just doing that. When I finally want to learn the song, I just find the lyrics online (usually from goo: bit.ly/aFc4H9) correct them as I see fit (add more kanji, correct wrong kanji, look up phrases I don’t know) and then add one line at a time in chunks of 10-20 at a time in no certain order.
After about a week~month you’ll be listening to the song and magically hear all those phrases you put in SRS jump out at you… *and* understand it 🙂
I’ve been trying to figure out the best way to memorize foreign song lyrics and couldn’t really figure out a good way to do this with an SRS, but this card setup has potential. Thanks, Khatz!
Dammit, was just about to write a post on that very subject! I see some other people have recommended Audacity – can’t do better than free 🙂 also, you can use Audacity to rip audio using a two ended 3.5mm jack since I don’t know that it still works for soundcard ripping
for anyone who likes the kind of j-rap I do, this video makes doing this really easy because the lyrics are typed out in easily legible black and white ARE the video, anyway check out SEEDA Answer:
www.youtube.com/watch?v=cLONNnYk3nM&feature=related
I’m on Loonix and I use mp3splt for this. I think there’s also a GUI for it, but I think this program’s command line interface is actually more comfortable to use.
Here’s how I’ve split my copy of the Paprika main theme (which you can download for free here) into 15-second chunks with 5 seconds overlap:
mp3splt -f -t 0.10 -O 0.05 -d byakkoyaSplits 白虎野の娘.mp3
(I’ve SRS’d the command itself, too)
@Eric: thanks a ton! 😀
Khatzumoto, has anyone told you that your new サクラ banner is awesome? Because it is!
ITunes will let you split files, just press cmd+i (ctrl+i on windows) then somewhere under one of the headings it’ll let you choose starts and finish times of the track, change it to what you want and then click okay. Once that’s done simple right click on the track and choose “create MP3 version (or AAC depending on how you have it set-up” from the menu. Beats having to download any software.
@Kendo、ありがとう。
@Aleph0Null: Thanks for the idea of using kana on the front… I actually struggle quite a bit with Kanji that are on the front of normal sentences as well and try to write them out without looking when I’ve seen the sentence so dang much that I just know what words will be there. Only problem would be differentiating from words that are normally written in kana quickly, perhaps using parens would help… onward!
I’ve been SRSing lyrics for a while now.
I generally delete the cards after the interval is more than a month or so, since I will probably listen to the song more often than that.
Katz, the new banners are awesome. I just gotta say it.
Great to see a post about music–never considered using SRS to learn songs; I mostly just read the lyrics once and then listen to the song a lot.
Some newish mp3 players, or at least the one I got in China (for about $25), can display the lyrics along with the song if you get an LRC file with the same filename as the song.
One of my favourite SRS moments is hearing a word I added from another source in a song.
This site is great for Japanese song lyrics:
j-lyric.net/
For me when I use Anki for song lyrics, it complains about leeches a lot. I guess that because a lot of the words are new, but since I want to know the song I just don’t let it suspend leeches.
lol (this is farce btw)
“As white people age, they start to feel more and more angry with their parents for raising them in a monolingual home. At some point in their lives, most white people attempt to learn a second language and are generally unable to get past ordering in a restaurant or over-pronouncing a few key words. This failure is not attributed to their lack of effort, but rather their parents who didn’t teach them a new language during their formative years.[emphasis mine]”
stuffwhitepeoplelike.com/2008/02/28/78-multilingual-children/
Spaced Repetition Mentioned in JET Japanese learning materials. Made me smile.
tinyurl.com/yhs7ck6
Khatz, thanks for bringing this interesting idea to our attention, but for us who like hip-hop, have you experimented any with this type of genre of music? I know with J-POP, 30 seconds of music time will get you a line or two in 30 sec., but with hip-hop, rappers can end up with 5-10 verses in 30 seconds, which seems like it could be tough and tiring from the start.
I’m about to try this idea with some hip-hop, so I just wanted to know if you could recommend any advice that would be relevant to this genre of music.
stuffwhitepeoplelike.com/2008/02/07/58-japan/
“[You] have to be careful about how much you like Japan. If you know how to speak Japanese, you kind of ruin it for everyone else.”
I think I need to apologize to some people I know. Some of them look longingly at the sheet I use to write missed SRS bits, which is filled with Japanese of varying sizes and arrangements. Every once in a while one asks me what something says, and when I tell them they’re like “Nice,” but now I know what they really meant was “You’ve ruined Japan for me.” My bad.
lol drewskie what’s sad is how true that actually is… They probably walk away mumbling, “He must have gotten to do a student exchange” or “His parents must have lived in Japan during his formative years,” but definitely not, “Wow, he must have worked really hard, I bet if I worked hard I could learn some Japanese too.”
Its like writing a novel. I’ve written four (really, really bad ones) and what I always hear is, “I have a great idea for a novel, if I just had time to write it.” As if the magical time fairy lived in my apartment or something… Its not like I didnt have a full time job, school, social life, etc when I wrote those, and the fourth was much, much much better than the first.
Hey Kreva, did you see the link to the SEEDA video I posted? There are several of his songs up on youtube with the full lyrics displayed really cleanly which should at least take some of the pain and suffering out of setting up the cards, even if you just use it to check against lyrics you cut and paste form elsewhere
Learning Japanese is easy people.
You let the SRS do the work, you let Japanese material do the work. All you do is show up.
Simple isn’t it?
Doesn’t even require thought. You don’t become fluent in Japanese, you become used to it. You become one with the language and the rest works itself out people.
@kendo
Getting lyrics wasn’t really where I was getting at, just the amount of words a rapper can rap inside of 30 seconds compared to that of a J-POP singer and whether it would be too much for a single SRS card, but all of that is irrelevant now. I experimented with some hip-hop already and even at 30 seconds a clip, its still a relatively small amount of text for each card.
Thanks again for this idea, Khatz. Applying this method to J-Hip-Hop is A-OK. 😀
ah, well glad that works well since that’s really all I listen to…I’ve noticed that japanese rappers that really go off are still pretty slow compared to someone like Do or Die, or Twista…
Also, I think khatz was implying that you would have multiple cards per audio clip to focus on short, discrete bits of info…So, even if you are listening to a song with a good bit of lyrics per sec, you can still break up what your focus on per card actually is
I use Audio Hijack Pro. I can rip sound from any source on my computer: Firefox, iTunes, DVD Player… it’s pretty awesome! While I watch movies or anime on my computer, it’s automatically recording it and creating new files from it every 5 minutes. You can adjust how much time lapses before it creates new files.
It also has a split feature which is particularly useful for music. From what I understand, Khatz, the way you do it seems like the audio created for your SRS are random 15-30 second clips. Which is why you suggest 5 seconds of overlap. The split feature in Audio Hijack is a lot more convenient, though, I think. Whenever I click “Split”, it creates a new file from that exact point. So as I’m listening to/hijacking a song from my iTunes, I click “Split” right before a lyric starts and right after it ends. This way I can have an audio file of a line (or entire stanza or whatever) in a perfectly packaged form. Now when I SRS with this audio file, I know it’s going to get right to the point of what I want to learn and also end promptly.
I just have the free demo version of Audio Hijack Pro which is all you will ever need if you’re using it for SRSing. The free version seems to give you all the features of the regular version, but it limits your hijacking to ten minutes at a time. Plenty of time for this kind of thing.
@KREVA my hiphop-rap-syllable-dense-song-learning-method points.
mix, match, etc these two
* sing or hum with, the whole song
* loop small sections 10-30 second bits and match the sounds
always do these two
* do it every day
* leave it when it’s fun
about lyrics:
Me, I can’t read first and sing every yet. So I just sing. Make the sounds of the song a reflex and then I can read and sing. I use the lyrics to help me clarify sounds, but sometimes they’re no help (typos, creative kanji readings).
Also, if you’re struggling please don’t blame the song or yourself…though that’s certainly easy, hehe. It’s the love of the idea inside you that inspires the energy to pursue your dream (like Peter Pan’s happy thought). So rather than struggle, think, feel, use intuition/imagination. Separate your mind from the illusion of reality and find the idea(s) that produces the love.
Happy hummin’ y’all
Has anyone tried this method with audiobooks (or soap operas)?
I did this with Singuila’s Ma Nature today. It was actually really fun. But he’s right, if it seems boring to you then stop.
I used Youtube downloader to rip the youtube video, then converted it to a mp3 with the youtube downloader. Then I used audacity to trim each part, and put it into anki with the audio and whatnot with about 3-4 lines per card. It seems like a lot, but really It took me maybe 45 minutes, and I stopped to take a phone call. Once you know the software (which takes 5 minutes) you can knock it out.
Now, I just add the full song to my mp3 with the other things and then when I review, it will be familiar.
I am thinking about using this with soap operas with a lot of dialog.
As you say, it’s possible to use Audacity to trim each part and put audio (L2) in the front card and subtitles (L2) in the back card. Using SRS in this way can be a sort of audio recognition training in your L2.
Microsoft Movie maker does the job of splitting songs perfectly to me… And its easy and simpo
Does anyone know what keywords I have to search for to get to Japanese websites for lyrics of Japanese Songs?
I would love to learn 素晴らしき日常 by 高橋優, but I cannot catch every word in the song or read/reproduce the respective kanji (although displayed prominently in the 間中-titles, thanks a lot).
www.youtube.com/watch?v=p5mw8_oG_fM&videos=vDAuLWY8Fck
So I guess it would be less time-consuming for this and following songs to have the correct lyrics and study from there…
The word you want is 歌詞｢かし｣. Song name Artist 歌詞 usually does the trick.
good information,
there you can use one of best method.
1.Open www.mp3converter.net
2.Past the url of songs you want to convert it.
3.click on convert and download it.