[hts-users:01750] A question about MSD-HMM
- Subject: [hts-users:01750] A question about MSD-HMM
- From: "Xingchi Xian" <yeahxxc@xxxxxxxxx>
- Date: Wed, 29 Oct 2008 21:00:15 +0800
- Delivered-to: hts-users@xxxxxxxxxxxxxxx
- Dkim-signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=gmail.com; s=gamma; h=domainkey-signature:received:received:message-id:date:from:to :subject:mime-version:content-type; bh=9mAM1HSKWusFQiB0mO2cPAicf+RynPOBdoMmEdaRAdQ=; b=vZq+q7gqU9kYW/9lFMruejIfX96ZkTrjgS4yoWcRXVWYyZjO3lmDMX7uqlM2ksIMha Y724bk3ROU20RAAhDRm5HsVjz7M9MoX5T6f4O/NCaQRi/xjJn98IxP10lCJLHcumLkVt WPPSLWKAaejAKZWsZTTxrrMsXC8QtgXOJjxHg=
- Domainkey-signature: a=rsa-sha1; c=nofws; d=gmail.com; s=gamma; h=message-id:date:from:to:subject:mime-version:content-type; b=hOtMrhUYz38RKpUisSuZgIyEramaIna+js4N0KHMhlHA0zUlF+PzK+kxMAfo8lOX9J pHxI/oeWeNVpcncIUM8VymUYiMZply5Usn9gZsijCH3oFSjhXf56YYau2OcNghq6/nnA nh70IMhR9GKKUFBtqVSwv2NakakrHZyAHVbPc=
Hi all,
I have a fundamental question about the MSD-HMM. In the f0 modeling case, a MSD-HMM state could output a unvoiced observation in a certain probability and a voiced observation in a certain probability density respectively. What perplexes me is the mathematical comparability between probability and probability density. Subsequently, this arouses two concrete question as follows.
1. How could we tell which is bigger, the likelihood of a state generating a voiced value represented by probability density which can be larger than 1(even it has been multiplied with the space weight), and the probability of the unvoiced symbol which is definitely less than 1. Just by space weight?
2. What's the mathematical meaning of the likelihood of a state sequence generating a f0 observation sequence, namely
(probabilty) x (probability density) x (probability density) x (probabilty) ...
Though MSD-HMM is implemented successfully, I am still glad to hear a explanation about these or what I have misunderstood.
Thanks and regards,
Xian Xingchi