Training speaker recognition system on LIBRISPEECH database

[Ahilan Kanagasundram]

Hello,

I'm trining to develop a speaker recognition system using librispeech database as this is the only database which can be used for commercial purposes. Could you please advise me whether this database can be used for speaker recogntion development. Do we have multiple sessions / speaker data in order to PLDA model?

I look forward to hearing from you.

Many thanks and best regards,
K.Ahilan

[Daniel Povey]

You could try but I'm not sure that it will work that great, since I
think we don't have any data (or much data) from the same speakers
with different recording conditions.
Dan

> --
> You received this message because you are subscribed to the Google Groups
> "kaldi-help" group.
> To unsubscribe from this group and stop receiving emails from it, send an
> email to kaldi-help+...@googlegroups.com.
> For more options, visit https://groups.google.com/d/optout.

[Ahilan Kanagasundram]

Thanks Dan. Yes, that's true, if we don't have same speakers from different recoding condition, it's hard to train better PLDA model.

Many thanks and best regards,
K.Ahilan