Quantcast

[QtMoko] MelodiQ - Music Recognition App

classic Classic list List threaded Threaded
5 messages Options
ANT
Reply | Threaded
Open this post in threaded view
|  
Report Content as Inappropriate

[QtMoko] MelodiQ - Music Recognition App

ANT
Hello,

I recently found an interesting online music recognition service [1] that allows you to upload a short (e.g. 10-15 sec) audio fragment (in almost any popular format) directly to the project's web page and get the information about the track title, artist's name, album, etc. It recognizes even low-quality microphone records, so I wrote a Shazam[2]-like GUI for it.

Please meet MelodiQ - a music recognition app that can be really useful in some cases. Source codes are hosted on GitHub [3]. The demonstration video is available on Youtube [4]. Note that you should place the phone very close to the source of sound while recording a sample. It is also necessary to solve the CAPTCHA to get the recognition results.

The database of this service is limited so it can't recognize absolutely each existing songs.
Don't forget to shock your friends who never heard about such kind of programs.

[1] http://audiotag.info/
[2] http://www.shazam.com/
[3] http://github.com/Sektor/melodiq
[4] http://www.youtube.com/watch?v=6dOT2yjkvwI

Have a Fun!
Anton
ANT
Reply | Threaded
Open this post in threaded view
|  
Report Content as Inappropriate

Re: [QtMoko] MelodiQ - Music Recognition App

ANT
BTW, I used the following commands to capture the microphone's output:
 $ alsactl -f /usr/share/openmoko/scenarios/voip-handset.state restore
 $ arecord -D plughw:0,0 -r 11025 sample1.wav
 Ctrl+C
 $ alsactl -f /usr/share/openmoko/scenarios/stereoout.state restore
but I noticed that recorded sound was _very_ quiet, so I decided to add one more command to increase volume level:
 $ sox -v 15 sample1.wav sample2.wav
Please e-mail me if you know a better solution for recording music with Freerunner's mic.

Cheers,
Anton
ANT
Reply | Threaded
Open this post in threaded view
|  
Report Content as Inappropriate

Re: [QtMoko] MelodiQ - Music Recognition App

ANT
Youtube video has moved. Here is a new link:
http://www.youtube.com/watch?v=b4LWaOsWY_E

Cheers,
Anton
Reply | Threaded
Open this post in threaded view
|  
Report Content as Inappropriate

Re: [QtMoko] MelodiQ - Music Recognition App

Alex Samorukov-2
In reply to this post by ANT
On 09/19/2010 03:31 PM, ANT wrote:

> I recently found an interesting online music recognition service [1] that
> allows you to upload a short (e.g. 10-15 sec) audio fragment (in almost any
> popular format) directly to the project's web page and get the information
> about the track title, artist's name, album, etc. It recognizes even
> low-quality microphone records, so I wrote a Shazam[2]-like GUI for it.
>
> Please meet MelodiQ - a music recognition app that can be really useful in
> some cases. Source codes are hosted on GitHub [3]. The demonstration video
> is available on Youtube [4]. Note that you should place the phone very close
> to the source of sound while recording a sample. It is also necessary to
> solve the CAPTCHA to get the recognition results.
>
>
>    
Thank you. I tried to test it, but for some reason it fail to recognize
any sample :( I tried radio and also some songs on a notebook, but no
luck. Could you please add "preview" mode before sending data to the
server? May be my mic settings are wrong?


_______________________________________________
Openmoko community mailing list
[hidden email]
http://lists.openmoko.org/mailman/listinfo/community
ANT
Reply | Threaded
Open this post in threaded view
|  
Report Content as Inappropriate

Re: [QtMoko] MelodiQ - Music Recognition App

ANT
On 09/19/2010 11:44 PM, Alex Samorukov-2 wrote:
>Could you please add "preview" mode before sending data to the
>server? May be my mic settings are wrong?

Recorded sample is stored in /tmp/melodiq-sample2.wav. You can try to play it right after recording. If this audio is recognizable by human and has sufficient length and the database contains a fingerprint of this song then it should be recognized.

Anton
Loading...