GSoC audio fingerprinting for Android

classic Classic list List threaded Threaded
7 messages Options
Reply | Threaded
Open this post in threaded view
|

GSoC audio fingerprinting for Android

tom kel
Dear MusicBrainz,

I am a student at the University of Kentucky, first year. I am very interested in joining development for the summer of 2012 (and beyond!).

This project will largely be based around 3 steps:

1. Edit acoustid-server to support fingerprint lookups of short audio segments.

2. Reimplement or port chromaprint to Java (Android).

3. Upgrade musicbrainz-server to support AcoustID

The end result will be able to fingerprint a short (30-second) clip recorded on your Android phone.

Please provide your thoughts on this project; I am very eager to work with you on this. MusicBrainz already exists for so much good and I hope to further that good to be in everyday use by the existence of this project. :)

        -Tommy Kelly

_______________________________________________
MusicBrainz-devel mailing list
[hidden email]
http://lists.musicbrainz.org/mailman/listinfo/musicbrainz-devel
Reply | Threaded
Open this post in threaded view
|

Re: GSoC audio fingerprinting for Android

Paul Taylor-2
On 26/03/2012 03:48, tom kel wrote:

> Dear MusicBrainz,
>
> I am a student at the University of Kentucky, first year. I am very
> interested in joining development for the summer of 2012 (and beyond!).
>
> This project will largely be based around 3 steps:
>
> 1. Edit acoustid-server to support fingerprint lookups of short audio
> segments.
>
> 2. Reimplement or port chromaprint to Java (Android).
>
> 3. Upgrade musicbrainz-server to support AcoustID
>
> The end result will be able to fingerprint a short (30-second) clip
> recorded on your Android phone.
>
> Please provide your thoughts on this project; I am very eager to work
> with you on this. MusicBrainz already exists for so much good and I
> hope to further that good to be in everyday use by the existence of
> this project. :)
>
>         -Tommy Kelly
>
I think just step 2 would be a project in itself, and a very useful one
indeed.

Paul

_______________________________________________
MusicBrainz-devel mailing list
[hidden email]
http://lists.musicbrainz.org/mailman/listinfo/musicbrainz-devel
Reply | Threaded
Open this post in threaded view
|

Re: GSoC audio fingerprinting for Android

Lukáš Lalinský
On Mon, Mar 26, 2012 at 9:49 AM, Paul Taylor <[hidden email]> wrote:
> I think just step 2 would be a project in itself, and a very useful one
> indeed.

Yep, I agree. Implementing a Shazam-like lookups with Acoustid is
pretty much impossible, but many people would benefit from a native
Java version of Chromaprint. Not sure if MB would want that as their
GSoC project though.

Lukas

_______________________________________________
MusicBrainz-devel mailing list
[hidden email]
http://lists.musicbrainz.org/mailman/listinfo/musicbrainz-devel
Reply | Threaded
Open this post in threaded view
|

Re: GSoC audio fingerprinting for Android

tom kel
Why is it impossible? If short audio lookups are implemented, and possibly adapted for noise while recording, is it still impossible?

Tommy

"Lukáš Lalinský" <[hidden email]> wrote:

>On Mon, Mar 26, 2012 at 9:49 AM, Paul Taylor <[hidden email]>
>wrote:
>> I think just step 2 would be a project in itself, and a very useful
>one
>> indeed.
>
>Yep, I agree. Implementing a Shazam-like lookups with Acoustid is
>pretty much impossible, but many people would benefit from a native
>Java version of Chromaprint. Not sure if MB would want that as their
>GSoC project though.
>
>Lukas
>
>_______________________________________________
>MusicBrainz-devel mailing list
>[hidden email]
>http://lists.musicbrainz.org/mailman/listinfo/musicbrainz-devel


_______________________________________________
MusicBrainz-devel mailing list
[hidden email]
http://lists.musicbrainz.org/mailman/listinfo/musicbrainz-devel
Reply | Threaded
Open this post in threaded view
|

Re: GSoC audio fingerprinting for Android

Lukáš Lalinský
On Tue, Mar 27, 2012 at 5:02 PM, Tommy <[hidden email]> wrote:
> Why is it impossible? If short audio lookups are implemented, and possibly adapted for noise while recording, is it still impossible?

Because you can't adapt it to noisy audio without completely
redesigning it. The way the hashes are calculated makes it impossible
to identify anything that's not almost identical, with the advantage
that the fingerprints are smaller and near-identical lookups are
faster.

If you implement the short audio lookups, it would be only useful for
things like tracklisting an internet radio audio stream.

Lukas

_______________________________________________
MusicBrainz-devel mailing list
[hidden email]
http://lists.musicbrainz.org/mailman/listinfo/musicbrainz-devel
Reply | Threaded
Open this post in threaded view
|

Re: GSoC audio fingerprinting for Android

tom kel


2012/3/27 Lukáš Lalinský <[hidden email]>
On Tue, Mar 27, 2012 at 5:02 PM, Tommy <[hidden email]> wrote:
> Why is it impossible? If short audio lookups are implemented, and possibly adapted for noise while recording, is it still impossible?

Because you can't adapt it to noisy audio without completely
redesigning it. The way the hashes are calculated makes it impossible
to identify anything that's not almost identical, with the advantage
that the fingerprints are smaller and near-identical lookups are
faster. 

If you implement the short audio lookups, it would be only useful for
things like tracklisting an internet radio audio stream.

Lukas

Do you have any suggestions for a new system that would be able to identify audio recorded on a phone?

Why not make a system such as this one of your goals? It would do much to increase public knowledge of MusicBrainz, we'd gain more contributors, thus better collections of data etc.

Tommy

_______________________________________________
MusicBrainz-devel mailing list
[hidden email]
http://lists.musicbrainz.org/mailman/listinfo/musicbrainz-devel
Reply | Threaded
Open this post in threaded view
|

Re: GSoC audio fingerprinting for Android

Ben Chobot
I believe when we were looking for a replacement for PUIDs this was examined and determined to be full of patents. 

On Mar 27, 2012, at 2:40 PM, tom kel wrote:

Why not make a system such as this one of your goals? It would do much to increase public knowledge of MusicBrainz, we'd gain more contributors, thus better collections of data etc.


_______________________________________________
MusicBrainz-devel mailing list
[hidden email]
http://lists.musicbrainz.org/mailman/listinfo/musicbrainz-devel