Reply by jon222 April 19, 20072007-04-19
>On Apr 18, 6:55 am, "jon222" <s...@kis.fri.uniza.sk> wrote: >> I need to use it for testing a method when concatenating two diphones
(in
>> the time domain), where the base frequency is not equivalent for both >> diphones, so in the area of concatenation to resample frequency of the >> first diphone to the next one. > >What do you mean by "base frequency"? Do you mean the fundamental >frequency of the voicing in each diphone, so you're trying to make a >smooth transition in the fundamental across the concatenation? > 1. Are you aware that changing the sampling rate like this will also >change the formant frequencies? That may be undesirable. > 2. There are other (perhaps easier) ways to modify the fundamental >frequency in the time domain. I suggest that you search for SOLA or >PSOLA. > >cheers, > jerry > >
Yes I mean the fundamental frequency of the voicing. 1. I'm aware of the formant changings, but still I have to test it, so I need to know how to gradually stretch it, so if there some Java code it will be helpful. 2. I'm going to test PSOLA methods too. J. _____________________________________ Do you know a company who employs DSP engineers? Is it already listed at http://dsprelated.com/employers.php ?
Reply by Jerry Wolf April 18, 20072007-04-18
On Apr 18, 6:55 am, "jon222" <s...@kis.fri.uniza.sk> wrote:
> I need to use it for testing a method when concatenating two diphones (in > the time domain), where the base frequency is not equivalent for both > diphones, so in the area of concatenation to resample frequency of the > first diphone to the next one.
What do you mean by "base frequency"? Do you mean the fundamental frequency of the voicing in each diphone, so you're trying to make a smooth transition in the fundamental across the concatenation? 1. Are you aware that changing the sampling rate like this will also change the formant frequencies? That may be undesirable. 2. There are other (perhaps easier) ways to modify the fundamental frequency in the time domain. I suggest that you search for SOLA or PSOLA. cheers, jerry
Reply by Jerry Avins April 18, 20072007-04-18
jon222 wrote:

   ...

> It should be the same steady sampling rate as the original signal, just > the new signal should gradually change the frequency (e.g. 100 - 200%), > with the term gradual resample I wanted rather to say gradual stretch or > gliding stretch.
Thanks for the response. I misunderstood what you wanted. I see that you already have some answer. Jerry -- Engineering is the art of making what you want from things you can get. &macr;&macr;&macr;&macr;&macr;&macr;&macr;&macr;&macr;&macr;&macr;&macr;&macr;&macr;&macr;&macr;&macr;&macr;&macr;&macr;&macr;&macr;&macr;&macr;&macr;&macr;&macr;&macr;&macr;&macr;&macr;&macr;&macr;&macr;&macr;&macr;&macr;&macr;&macr;&macr;&macr;&macr;&macr;&macr;&macr;&macr;&macr;&macr;&macr;&macr;&macr;&macr;&macr;&macr;&macr;&macr;&macr;&macr;&macr;&macr;&macr;&macr;&macr;&macr;&macr;&macr;&macr;&macr;&macr;&macr;&macr;
Reply by jon222 April 18, 20072007-04-18
>jon222 wrote: >> Does anyone have Java or C code for a gradual resampling. I already
have
>> implementation of Nyquist&#2013266070;Shannon sampling theorem, where I can
resample
>> to any ratio. I know to calculate the amplitude of the signal at any
time.
>> Now I need to know at which time to take the amplitude to achieve the >> gradual resampling e.g. from 100% - 200%. It means at the beginning it
is
>> the same signal and gradually to the end the length is stretched to
200%.
>> >> Thanks >> J. >> > >You can do this with the excellent libsamplerate >(http://www.mega-nerd.com/SRC/), or you can write your own >implementation of bandlimited interpolation[1] which gradually changes >the sample rate. > >Why do you want to do this? > >Cheers > >Marc > >[1] http://ccrma-www.stanford.edu/~jos/resample/ >
I need to use it for testing a method when concatenating two diphones (in the time domain), where the base frequency is not equivalent for both diphones, so in the area of concatenation to resample frequency of the first diphone to the next one. Does exist any Java package with dsp methods? J. _____________________________________ Do you know a company who employs DSP engineers? Is it already listed at http://dsprelated.com/employers.php ?
Reply by Marc Brooker April 18, 20072007-04-18
jon222 wrote:
> Does anyone have Java or C code for a gradual resampling. I already have > implementation of Nyquist&#2013266070;Shannon sampling theorem, where I can resample > to any ratio. I know to calculate the amplitude of the signal at any time. > Now I need to know at which time to take the amplitude to achieve the > gradual resampling e.g. from 100% - 200%. It means at the beginning it is > the same signal and gradually to the end the length is stretched to 200%. > > Thanks > J. >
You can do this with the excellent libsamplerate (http://www.mega-nerd.com/SRC/), or you can write your own implementation of bandlimited interpolation[1] which gradually changes the sample rate. Why do you want to do this? Cheers Marc [1] http://ccrma-www.stanford.edu/~jos/resample/
Reply by jon222 April 18, 20072007-04-18
>jon222 wrote: >> Does anyone have Java or C code for a gradual resampling. I already
have
>> implementation of Nyquist&#65533;Shannon sampling theorem, where I can
resample
>> to any ratio. I know to calculate the amplitude of the signal at any
time.
>> Now I need to know at which time to take the amplitude to achieve the >> gradual resampling e.g. from 100% - 200%. It means at the beginning it
is
>> the same signal and gradually to the end the length is stretched to
200%.
> >Most signal processing algorithms depend on a steady sample rate. What >will you do with your chirped signal? > >Jerry >-- >Engineering is the art of making what you want from things you can get. >&macr;&macr;&macr;&macr;&macr;&macr;&macr;&macr;&macr;&macr;&macr;&macr;&macr;&macr;&macr;&macr;&macr;&macr;&macr;&macr;&macr;&macr;&macr;&macr;&macr;&macr;&macr;&macr;&macr;&macr;&macr;&macr;&macr;&macr;&macr;&macr;&macr;&macr;&macr;&macr;&macr;&macr;&macr;&macr;&macr;&macr;&macr;&macr;&macr;&macr;&macr;&macr;&macr;&macr;&macr;&macr;&macr;&macr;&macr;&macr;&macr;&macr;&macr;&macr;&macr;&macr;&macr;&macr;&macr;&macr;&macr; >
It should be the same steady sampling rate as the original signal, just the new signal should gradually change the frequency (e.g. 100 - 200%), with the term gradual resample I wanted rather to say gradual stretch or gliding stretch. J. _____________________________________ Do you know a company who employs DSP engineers? Is it already listed at http://dsprelated.com/employers.php ?
Reply by Jerry Avins April 17, 20072007-04-17
jon222 wrote:
> Does anyone have Java or C code for a gradual resampling. I already have > implementation of Nyquist&#65533;Shannon sampling theorem, where I can resample > to any ratio. I know to calculate the amplitude of the signal at any time. > Now I need to know at which time to take the amplitude to achieve the > gradual resampling e.g. from 100% - 200%. It means at the beginning it is > the same signal and gradually to the end the length is stretched to 200%.
Most signal processing algorithms depend on a steady sample rate. What will you do with your chirped signal? Jerry -- Engineering is the art of making what you want from things you can get. &macr;&macr;&macr;&macr;&macr;&macr;&macr;&macr;&macr;&macr;&macr;&macr;&macr;&macr;&macr;&macr;&macr;&macr;&macr;&macr;&macr;&macr;&macr;&macr;&macr;&macr;&macr;&macr;&macr;&macr;&macr;&macr;&macr;&macr;&macr;&macr;&macr;&macr;&macr;&macr;&macr;&macr;&macr;&macr;&macr;&macr;&macr;&macr;&macr;&macr;&macr;&macr;&macr;&macr;&macr;&macr;&macr;&macr;&macr;&macr;&macr;&macr;&macr;&macr;&macr;&macr;&macr;&macr;&macr;&macr;&macr;
Reply by jon222 April 17, 20072007-04-17
Does anyone have Java or C code for a gradual resampling. I already have
implementation of Nyquist&#2013266070;Shannon sampling theorem, where I can resample
to any ratio. I know to calculate the amplitude of the signal at any time.
Now I need to know at which time to take the amplitude to achieve the
gradual resampling e.g. from 100% - 200%. It means at the beginning it is
the same signal and gradually to the end the length is stretched to 200%.

Thanks
J.



_____________________________________
Do you know a company who employs DSP engineers?  
Is it already listed at http://dsprelated.com/employers.php ?