Each line in the gauden_counts file(printable form) contains 36 pairs of floating point numbers and 1 float. As I read from this forum and other references, the pairs are counts for mean and variance respectively. Each line is for one gaussian and there are 32 gaussians for each senone. The last number in the array is the denominator used for normalization.
Is this correct?
I am trying to take an existing mean / variance file and apply the calculations based on one word (observation sequence) to see the changes in the parameters. The calculated values are different. That's why I need this clarification.
The formulas I use for calculation of mean and variance are given in the attached files.
I am back with the same question again. Pl pardon me if it is very trivial. But, I'm stuck without this being answered.
My gauden_counts file's one entry is like this:
The entries are adjustment counts for means and variances. Then, there have to be 36 pairs. But, here there are 36 pairs and one float. This float value is the denominator that is used for calculating means and variances from these counts (as per the formula I have attached in the previous post).
My query is,
1. Is the above statement correct?
2. If yes, where is the denominator placed in the array - beginning or end. Or in other words, the delimiter for the entries in gauden_counts is a space or a comma?
Thank you.
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
Each line in the gauden_counts file(printable form) contains 36 pairs of floating point numbers and 1 float. As I read from this forum and other references, the pairs are counts for mean and variance respectively. Each line is for one gaussian and there are 32 gaussians for each senone. The last number in the array is the denominator used for normalization.
Is this correct?
I am trying to take an existing mean / variance file and apply the calculations based on one word (observation sequence) to see the changes in the parameters. The calculated values are different. That's why I need this clarification.
The formulas I use for calculation of mean and variance are given in the attached files.
Thank you.
Hello,
I am back with the same question again. Pl pardon me if it is very trivial. But, I'm stuck without this being answered.
My gauden_counts file's one entry is like this:
The entries are adjustment counts for means and variances. Then, there have to be 36 pairs. But, here there are 36 pairs and one float. This float value is the denominator that is used for calculating means and variances from these counts (as per the formula I have attached in the previous post).
My query is,
1. Is the above statement correct?
2. If yes, where is the denominator placed in the array - beginning or end. Or in other words, the delimiter for the entries in gauden_counts is a space or a comma?
Thank you.