bob.spear applicable to speaker diarization?

[Takafumi Koshinaka]

Hello,

I am an engineer who became a user of bob.spear after I found it in ICASSP14.

Now I am wondering if there are any use cases of bob.spear in i-vector-based speaker diarization.

I think typical use cases so far focus on speaker recognition, where they assume that each audio file contains only one speaker.

It might be great if anyone has ever tried i-vector extraction for each utterance in an audio file which contains two or more speakers.

I would appreciate any ideas you can provide.  Thanks in advance for your kindness and support.

[Elie Khoury]

Hello,

Thank you for your message. Indeed, at Idiap, we tried speaker diarization (both segmentation and clustering) within the bob.spear framework. But unfortunately the code is not public yet. We will keep you updated in case it turns public.

Best regards,

--Sent from my mobile device--

--
-- You received this message because you are subscribed to the Google Groups bob-devel group. To post to this group, send email to bob-...@googlegroups.com. To unsubscribe from this group, send email to bob-devel+...@googlegroups.com. For more options, visit this group at https://groups.google.com/d/forum/bob-devel or directly the project website at http://idiap.github.com/bob/
---
You received this message because you are subscribed to the Google Groups "bob-devel" group.
To unsubscribe from this group and stop receiving emails from it, send an email to bob-devel+...@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

-- 
--------------------
Elie Khoury
Post Doctorant
Biometric Person Recognition Group
Idiap Research Institute
Switzerland 
CH : +41 27 721 77 23

[Takafumi Koshinaka]

Dear Dr. Khoury,

 

Thank you!  I look forward to your code getting available.
And I would appreciate if you could give me any suggestion for how to feed segmentation results of my own speaker segmenter into spear's i-vector extractor.
If there is any possible way, I would like to try.

 

Best regards,

 

----
Takafumi KOSHINAKA, Ph.D.
Information and Media Processing Labs., NEC Corporation

 

2014年9月16日火曜日 1時09分00秒 UTC+9 Elie Khoury:

[manish...@quantiphi.com]

Hi Dr. Khoury,

can you please update us regarding the dirization project in bob.spear also it would be really helpful if you could give some suggestions regarding the segment level i-vector extraction part.

Thanks & Regards

Manish Kumar

India: +91 7411723707