Displaying 8 results from an estimated 8 matches for "hypergeom_gain".
2009 Jul 22
2
A technical question about the speex preprocessor.
I got the approximation from a Google book:
http://books.google.com/books?id=2CAqsF-RebgC&pg=PA385
Page 392, formula (10.33)
Using this formula, you're right, hypergeom_gain() would *not* converge
to 1 for large x, but would instead be gamma(1.25)/sqrt(sqrt(x)) which
would approach zero. Now if the formula for the hypergeometric gain were
instead gamma(1.5) * M(-.5;1;-x) / sqrt(x) that *would* approach 1, but
that's just me noodling around with the formula to g...
2009 Jul 22
2
A technical question about the speex preprocessor.
...b-a))*(x^-a)
Which for M(-.25;1;-x) is sqrt(sqrt(x))*1.10326
at x=10 this gives a value of 1.96191 (vs. what I think is the true
value of 1.97456).
The reason I've gotten into all this is I'm trying to vectorize with SSE
intrinsics some of the slower loops in the preprocessor, and the
hypergeom_gain function is the only thing stopping me from removing all
the branches in the loops. I don't know how critical the accuracy of the
function is to the performance of the preprocessor, but if that
aforementioned approximation was good enough for all the values of x, it
would really speed the l...
2009 Jul 22
2
A technical question about the speex preprocessor.
...hat it was the confluent hypergoemetric function, which I was
trying to evaluate using Kummer's equation, namely:
M(a;b;x) is the sum from n=0 to infinity of (a)n*x^n / (b)n*n!
where (a)n = a(a+1)(a+2) ... (a+n-1)
But when I use Kummer's equation, I don't get the values in the
"hypergeom_gain" table. Did you use a different solution to the
confluent hypergoemetric function when you created the table?
John Ridges
Jean-Marc Valin wrote:
> Hi John,
>
> M(;;) is the confluent hypergeometric function.
>
> Jean-Marc
>
> John Ridges a ?crit :
>
>> Hi,...
2009 Jul 23
0
A technical question about the speex preprocessor.
...doc and the code, the code is the one likely to be
correct.
Jean-Marc
John Ridges a ?crit :
> I got the approximation from a Google book:
>
> http://books.google.com/books?id=2CAqsF-RebgC&pg=PA385
>
> Page 392, formula (10.33)
>
> Using this formula, you're right, hypergeom_gain() would *not* converge
> to 1 for large x, but would instead be gamma(1.25)/sqrt(sqrt(x)) which
> would approach zero. Now if the formula for the hypergeometric gain were
> instead gamma(1.5) * M(-.5;1;-x) / sqrt(x) that *would* approach 1, but
> that's just me noodling around with...
2009 Jul 21
2
A technical question about the speex preprocessor.
Hi,
I've been trying to re-create the table in the function "hypergeom_gain"
in preprocess.c, and I just simply can't get the same values. I get the
same value for the first element, so I know I'm computing gamma(1.25)^2
correctly, but I can't get the same numbers for M(-.25;1;-x), which I
assume is Kummer's function. Is it possible that the commen...
2009 Jul 22
0
A technical question about the speex preprocessor.
Something looks odd without your values (or the doc) because hypergeom_gain()
should really approach 1 as x goes to infinity. But in the end, an
approximation is probably OK because denoising is anything but an exact science
:-)
Jean-Marc
Quoting John Ridges <jridges at masque.com>:
> By my reckoning the confluent hypergoemetric functions should have the
>...
2009 Jul 22
0
A technical question about the speex preprocessor.
...oemetric function, which I was
> trying to evaluate using Kummer's equation, namely:
>
> M(a;b;x) is the sum from n=0 to infinity of (a)n*x^n / (b)n*n!
> where (a)n = a(a+1)(a+2) ... (a+n-1)
>
> But when I use Kummer's equation, I don't get the values in the
> "hypergeom_gain" table. Did you use a different solution to the
> confluent hypergoemetric function when you created the table?
>
> John Ridges
>
>
> Jean-Marc Valin wrote:
> > Hi John,
> >
> > M(;;) is the confluent hypergeometric function.
> >
> > Jean-Marc
&...
2009 Jun 30
3
Delays estimation in Speex algorithms
Speex tells me that the decoder is always 5 ms, but it says that the
encoder is 5 ms for NB, 8.9375 ms for WB, and 10.90625 ms for UWB. Is
there an extra frame of delay in the encoder that isn't otherwise
accounted for?
John Ridges
Jean-Marc Valin wrote:
> Quoting John Ridges <jridges at masque.com>:
>
>> I also need to know the precise delays from Speex but I used