About ALIZE and Recognito

[Travis Banger]

Out of frustration with ALIZE, I discovered Recognito (cool name :-)

I have not programmed in Java for a very long time, but am willing to reinstall everything (Eclipse? Netbeans?) for Recognito.

I am extremely puzzled about ALIZE and hope somebody here (Amaury?) can help clarify.

How come the ALIZE project was so enthusiastic for over a decade, but now you go to their forum (or mailing list) and it feels like somebody died? Perhaps it is related to a paper that describes a successful attack on ALIZE (the paper looks genuine, but it is apocryphal -- anonymous)?

Amaury: You recommended the book "Fundamentals of Speaker Recognition". How hard is it to follow?

Finally: There are 3 steps to ALIZE (building, running the tutorials and achieving a Speaker Identification/Verification), how come the 3rd. step seems to be ultra secret? People who know about that seem to be unwilling to share their knowledge, which is rare in an OSS project.

Regards and TIA

[Amaury Crickx]

I understand the frustration you had with ALIZE :)

I have not programmed in Java for a very long time, but am willing to reinstall everything (Eclipse? Netbeans?) for Recognito

Any IDE you kinda know how to find your way through is a good one 

I am extremely puzzled about ALIZE and hope somebody here (Amaury?) can help clarify.

How come the ALIZE project was so enthusiastic for over a decade, but now you go to their forum (or mailing list) and it feels like somebody died? Perhaps it is related to a paper that describes a successful attack on ALIZE (the paper looks genuine, but it is apocryphal -- anonymous)?

This is a university research project and I guess the guys just moved to something else... as you probably know, ALIZE is basically a platform for searchers in biometric identification. The goal is to make it "easy" to add new algorithms. The problem is that these are smart ppl but not very talented programmers...

Most probably, most of them now work for companies providing solutions around voice authentication... 

As for the attack, I don't think they really care... of course the system can be fooled, there aren't even safeguards for fraud detection... I remember there was a disclaimer on top of each file, saying something like "don't trust the results too much" 

 

 

Amaury: You recommended the book "Fundamentals of Speaker Recognition". How hard is it to follow?

Depends which part. I can't really follow most of the math notation but even skipping the formulas, I could read the text and understand what they're saying. Extremely instructive.

 

 

Finally: There are 3 steps to ALIZE (building, running the tutorials and achieving a Speaker Identification/Verification), how come the 3rd. step seems to be ultra secret? People who know about that seem to be unwilling to share their knowledge, which is rare in an OSS project.

I don't get how it can be kept as a secret when we have source code :-) explanations are a bit hard to follow I'm afraid, maths with imaginary numbers and advanced statistics :-) 

 

[JF Bonastre]

Hello
I am at the inititaive of ALIZE..
This message comes  a long time after your exchanges.
But I would like to understand what were your objectives with ALIZE and your deceptions.
I would like to know why you didn't obtain any answer...
Did you try to subscribve to dev-alize mail list ? To the linkedin group?
To contact me?

That's true, ALIZE was sleeping a little bit during the last years, due to a lack of time and money. More than 150 institutional users in the world but few contributors...
Don't mean that we didn't make progresses, just we don't post all on the evanescent website...
We (a set of french academic labs) are planning to relaunch very soon ALIZE with new stuff, including Python top programs
But we are not a company!! If you want a product based on ALIZE you could:
- make an agreement with us, with money...
- contact one of the companies who are making industrial products with ALIZE

Best
JF Bonastre
I

[Amaury Crickx]

Dear JF,

I humbly believe you should gleefully dismiss what 2 blokes were ranting at a while back. This forum is actually dead (was it ever alive or at least somewhat active?)

But I understand your frustration, we are indeed publicly criticising the work of talented researchers without actually reaching out to them for explanations and how-to's... and Google knows about it...

So, in an attempt to clarify my previous comments:

ALIZE is simply the best open source speaker recognition software available. But it's not meant for everyone and that is extremely frustrating.

The frustration we felt is similar to that of the illiterate person that would die to be able to read a particular book but simply can't... and it doesn't sound reasonable to ask some stranger to read the book from end to end...

Usually, when there is no active community it means the project is dead and might not resurrect... so why bother some busy scientist when all you want to do for the moment is give it a try during the next couple of hours, just to see what you can do with it. 

After you read big chunks of the documentation and parts of the code during a few hours, you realise you have no clue what this all means because most of the terminology is completely foreign to you... mmmeh, you feel both stupid for not getting it and angry because you know it could have been much more easy to deal with

I guess it might be ok like that, you don't have to make it easy for "illiterate" people, they're not your target audience, are they?... they'll be frustrated and rant in some corner of the internet, who cares :-)

Looking forward to the new ALIZE developments

Cheers

Amaury