[hts-users:02981] Re: diagonalization of covariance matrices

[Hui LIANG <tshlmail-hts@xxxxxxxxx>]

I've found out the answer to my first question in the archive of this mailing list. Any comments on the second point are still welcome and appreciated.

Best regards,
Hui LIANG



----- Original Message -----
From: Hui LIANG <tshlmail-hts@xxxxxxxxx>
To: "hts-users@xxxxxxxxxxxxxxx" <hts-users@xxxxxxxxxxxxxxx>
Cc: 
Sent: Wednesday, 3 August 2011, 15:08
Subject: [hts-users:02980] diagonalization of covariance matrices

Hello,

Could anyone confirm that when converting a model adapted by CMLLR transforms into an HTS engine, the resulting speaker-specific, full covariance matrices are diagonalized by HTS?

If so, I wonder whether there is any paper comparing the performance of synthesis with the original full covariance matrices and diagonalized ones? I am curious about the performance gap between the two cases.

Thank you very much in advance!

Best regards,
Hui LIANG