On Monday 07 August 2017 01:49:06 Fred van Stappen wrote:
> There is a buffer of float filled by the input of a sound card, or/and by a
> decoded audio file, or/and by a custom sine wave, etc...
> I would like to analyse that buffer (array of float = the samples).
> Can mse fft analyse other sources than from mse audio ? (huh, like for
> example from uos...)
You can use a "tfft" from component palette tab "Math" which has no MSEsignal 
interface. An example is here:

> Concretely, how to analyze the array of float (the samples) to get the
> frequencies that produce that array, what is the 1, 2, 3, 4 trick ?
What do you mean with "1, 2, 3, 4 trick"?

FFT means "Fast Fourier Transform":
an efficient special case of DFT:
which is a sampled form of the continuous Fourier-transform:

> Let's begin with something simple: a unique sine wave of 440 htz produced
> by a math function.
> If a array[0..1023] of float was filled, with samplerate 44100, by a
> sine-wave of 440 htz, if you do not know the freq of the sine wave (only
> the sample rate), how a fft can retreive the frequency of the signal ? How
> many array[0..1023] of float are needed to give a good result ?
> And to continue with something more complicated, if the buffer was filled
> by the input of a sound card (mic or aux, here a string of guitar), how to
> get/note the main frequency (for tuning if needed) ?
> I would like, for each guitar string, design the "ideal" sine wave (440 htz
> for la) in green, and design the live-sine wave of the guitar in red. 
> While tuning the string, when both green and red waves are equal, then it
> is tuned. But a [ +/- the note ] like classic tuners do is good too.
I don't know if it is possible to do precision tuning measurement with FFT. I 
fear not.

> And a spectrum/peak 16 bands for audio files/input mic would be very
> welcome too. I can do it with uos but I am courious how you propose to do
> it with mse-fft (and how fast it is).
The fft component of MSEgui uses the FFTW library as backend:
For realtime applications probably there are better solutions. For an audio 
spectrum display a filterbank probably is better suited because an FFT has a 
linear frequency scale and for audio one needs a logarithmic scale.

> OK, back to mse noisegen demo.
> What is scrollbar "crest" ?

It is visible in white noise mode. With crest factor 1 the sample values are 
in range -1..1, crest 2 -> -2..2 and so on.

> What king of average is done with checkbox "average" ?
It averages the spectrum values in order to show the "db/octave" 
characteristic of the noise.


Check out the vibrant tech community on one of the world's most
engaging tech sites, Slashdot.org! http://sdm.link/slashdot
mseide-msegui-talk mailing list

Reply via email to