AudioTag.Info - Music Recognition Robot, enters beta testing stage Options

[aradzish]

Dear All,

I'm glad to present to you free music recognition robot AudioTag.Info.

AudioTag.Info - is a free music-recognition service. It allows you to identify almost any unknown piece of music quickly and easily. Its use is very simple: you upload a short audio fragment or an entire song, the robot analyzes it and provides you with the information about the track title, artist name, album title, etc. Your audio fragment can be in almost any file format and of almost any quality (aurally recognizable, of course) — it can be an MP3 file downloaded from the Internet or a short recording made with your old tape recorder and stored as a low-quality WAV file.

The project enters beta-testing stage now.
Please visit, impress, test and report if you find any error or misbehavior.

Thanks.

This post has been edited by aradzish: Feb 21 2009, 21:08

[aradzish]

Sorry, a quick question. Do moderators for some reason think that this news isn't worth publishing it on the main page?
I'm just interested in letting people know about this project. The project is absolutely free so I though it might be interesting/useful for people.

This post has been edited by aradzish: Mar 3 2009, 20:37

[Qest]

Looks interesting. Did you write the fingerprinting algorithm yourself or is this using musicbrainz or something similar? By the way, love the "prove you are human", very witty.

[aradzish]

Qest,

thanks for your feedback.
All I can say about the algorithm is that it is proprietary (not MusicBrainz, etc). I cannot disclose details here, sorry.
The algorithm is much more precise than of MusicBrainz, Tunatic and others. You can play with audiotag robot and ensure in that by yourself.
I believe, FAQ there answers most of possible questions.
Though I'll be happy to answer every.

This post has been edited by aradzish: Mar 4 2009, 06:12

[aradzish]

Just for your information:
* you can see now "live statistics" at the bottom of the main page showing DB size and number of recognized music queries.
* also, a blog has been added allowing users to comment, discuss, leave feedback.

[aradzish]

Hello,

I thought this forum is for submission of news, and that after approval the news from here go to the main page,
but it seems that I was wrong and the news are actually filtered.

Really, I don't know why the information about AudioTag isn't worth placing at the main page.
Will appreciate to get some feedback from moderators.

Cheers.

[chornobyl]

I tried it once and didnt really liked
1 it gives me error 4-4 is not equal to 0, so now ama robot
2 it failed to recognize 30s flac recorded tru mic, and guessed the mp3 version, exacly this way acting winamps gracenote

[aradzish]

1 it gives me error 4-4 is not equal to 0, so now ama robot
2 it failed to recognize 30s flac recorded tru mic, and guessed the mp3 version, exacly this way acting winamps gracenote

1. I was reported about 4-4 captcha bug. I'll fix it.
2. There is no magic, but pure DSP processing - of course the algorithm is not absolute, but it has proven its efficacy on very noise recordings, phone recordings, etc. But... faults happen - your case is example.

[aradzish]

captcha bug is fixed.

Direct URL support is on the way.
(to identify audio from the web without the need to upload file).

[aradzish]

http://audiotag.info/blog/?p=11:

Now you can recognize files directly from the Web by specifying direct URL.

[Canar]

Usually how the News Submissions forum works is that if any moderator finds the submission worth posting, they post it. It's a pretty ad hoc process, generally-speaking.

No offense meant to your project, it seems interesting, but I'm not personally convinced it's front page material.

[aradzish]

Canar, thanks for your feedback.
Well, I understand the point.
Honestly, I was sure that this service is potentially interesting for the people because there is actually no such services available on the web. There are only services which require special software download and installation, while AudioTag works/available purely online. But well, if you do not find it interesting - no problem. I can update the audience about the project news here, in this topic. If you think this is improper - you can delete this topic and I'll not post about AudioTag on hydrogene anymore.

[Synthetic Soul]

I think that your service is most definitely "potentially interesting", and would certainly like to see future updates posted here.

Canar's point is simply that not every new product or service that is of interest to this board can be promoted to Validated News, we regularly move numerous threads from Submitted News to other forums.

I have actually already used your service three times: once to test (successful), once to identify a clip uploaded to this forum (fail, track was not English or Russian), and once in an attempt to tag a track which I had no idea of its heritage (successful, thanks).

[aradzish]

Thanks, Synthetic Soul, I'm glad to hear you find it useful and helpful.

There are many ideas on extending this service. I'll keep you all posted here if you're interested.

[uart]

Thanks aradzish, I just tested the site with a few 30 second samples and it picked them all correctly. That's very useful.

BTW. I really liked the fact that I didn't need to download any junk onto my computer to access it. I'm using a totally bland firefox install (without so much as even a flash plugin) and the site worked just perfectly. 100 points!

This post has been edited by uart: Mar 23 2009, 15:32

[aradzish]

Thanks for the positive fedback, uart.

By the way, you can upload even 5-10 seconds samples, they should be also recognizable.

[aradzish]

I keep you posted as promised.

Audiotag.info robot is now capable of dealing with several new file formats, in particular, .AMR (phones), .FLV (Flash Video), .MP4 and several more.

[dutch109]

Great service: free, powerful and easy to use.

I tested it with a sample ripped from a DVD, cut with Audacity and uploaded in Vorbis format, and it was successfully recognized.

However, your server seems to be overloaded due to its success, sometimes it takes 30sec to load the home page and it shows an error after the upload step.

[aradzish]

dutch109, thanks for the positive feedback. I'm glad you found it useful.

No, the "unavailability" of the service was not related to the high server load, just temporary problems with the provider.

[aradzish]

I've added upload progress bar. So, file uploading progress is now visualized.

[null-null-pi]

somehow it won't work for me. couldn'r recognize a 15 second sample of "dirty diana" by michael jackson recorded through my headset microphone. after that i tried madonna - material girl, also no success. i wonder if i'm doing something wrong.

edit: worked after i produced a 15 seconds mp3 file by directly cutting it (via mp3directcut). the quality of the files recorded through headset was approximately the same as what i get when i record something with my cellphone, so i'm not very confident about the robustness of the algorithm.

edit2: typo
edit3: correction again.

This post has been edited by null-null-pi: Apr 11 2009, 13:53

[aradzish]

null-null-pi, in general, there are no miracles. Transducing via air generally causes very serious damage to the audio that reduces its recognition chances significantly.
I can only say that with GSM recordings the algorithm managed mostly well.
However, it would be interesting to get/analyze your recordings used for tests (those you obtained by recording from your headset).
Can you please put them somewhere for me?

Addition: in general, criticism is very good but when you say that you're not sure in the robustness of this algorithm, can you please point me on any other service (free or commercial) available on the net which would recognize your test samples?

This post has been edited by aradzish: Apr 11 2009, 17:08

[null-null-pi]

null-null-pi, in general, there are no miracles. Transducing via air generally causes very serious damage to the audio that reduces its recognition chances significantly.
I can only say that with GSM recordings the algorithm managed mostly well.
However, it would be interesting to get/analyze your recordings used for tests (those you obtained by recording from your headset).
Can you please put them somewhere for me?

Addition: in general, criticism is very good but when you say that you're not sure in the robustness of this algorithm, can you please point me on any other service (free or commercial) available on the net which would recognize your test samples?

unfortunately i overwrote the recorded files out of sheer lazyness (untitled.mp3 was defaulted, so i just clicked "save" every time. didn't think i would still need them). but i can re-record the samples if you like.
i have to confess i didn't test ANY web based recognizer until today, so indeed i can't compare directly. but i once tested "TrackID" on my sonyericsson cellphone in a totally overcrowded disco and it identified the music at first attempt. that's why i was expecting this to work out, maybe i was a little bit overoptimistic. and sorry for my unprecise language, i'm not a native speaker so maybe sometimes i will choose wrong words and it may sound more aggressive than originally intended.

edit: typo again (and again)
edit3: not unnamed.mp3 but untitled.mp3

This post has been edited by null-null-pi: Apr 11 2009, 21:49

[aradzish]

null-null-pi, I'm not a native English speaker too, so that's fine And take it easy, you sounded very polite.
Please rerecord something and test on audiotag. If it won't be recognized - please send the recordings to me.
Thanks.

Cheers.

[null-null-pi]

i'll upload recordings as soon as i've got the time. maybe tomorrow.
out of curiosity i recorded some samples at a local disco tonight (again using my cellphone) and they also don't seem to be recognizable. but i have no clue what's behind the scene, it encodes to "amr" which i transcoded to mp3 via "mobile media converter". more to come, it's time to bang into the pillow now.