<simondex@yahoo.com> wrote in message
news:1124718668.266876.253260@g47g2000cwa.googlegroups.com...
> Hi, Everyone!
>
> If I have analyzed a sound and have its spectrogram, how can I utilize
> that information for sound synthesis?
>
> Thank You Very Much.
>
> Truly Yours, Simon Dexter
>
Simon,
Doing what you want can be very challenging. Here are some of the reasons:
- you may need to know the analysis bandwidth / resolution that will be
applied to the synthesized sound. That's because you need the record to be
long enough to not end up with a synthesized sound that is made up of
"lines" while the original was more continuous.
- you need to know what part of the spectrum is really broadband noise so
that it can be replicated adequately.
Take a look at speech synthesis. There's a huge body of literature. The
objective there was to compress speech and re-create it. So, there's a
short-term aspect to the methods - in contrast to the issue I mention above.
Fred
Reply by Rune Allnor●August 22, 20052005-08-22
simondex@yahoo.com skrev:
> Hi, Everyone!
>
> If I have analyzed a sound and have its spectrogram, how can I utilize
> that information for sound synthesis?
By computing the average spectrum and compute the Inverse
Fourier Transform?
This is an impossible question to answer, unless you provide
some more information:
- Do you have power spectrograms or magnitude/phase spectrograms?
Some times you ned the phase of the spectrum.
- Are the data stationary or are they transients?
Stationary data can to some extent be synthesized from spectra,
transients can be difficult.
- What is the application you have in mind? The constraints
imposed by the appliications can completely govern what
DSP methods can be considered. What can be used in off-line
processing need not be suitable for on-line processing.
Rune
Reply by ●August 22, 20052005-08-22
Hi, Everyone!
If I have analyzed a sound and have its spectrogram, how can I utilize
that information for sound synthesis?
Thank You Very Much.
Truly Yours, Simon Dexter