Sign in to follow this  
Followers 0
Snoozr

FFT for dummies

6 posts in this topic

Emphasis on the dummies part of that ;)

I was curious if anyone knows of software that can do the following - suppose I have a WAV file of me saying "Hi".  Suppose I would be curious to approximate this with, say, 10 sine waves.  So I load the wav file and the software will tell me it can be reproduced with "Sine wave 1: 1203 hz, length .7 seconds; Sine wave 2 ..."

I understand that a regular FFT will give me spikes at the right frequencies, but a.) I don't have regular software and b.) I was hoping to get more digestible output.

Thanks everyone.

Share this post


Link to post
Share on other sites

Ahhh resynthesis, one of my favourite toys! You forgot the partials' amplitude envelopes, and that's the bit you really want to know ;) That's why most wave editors do 3D FFT... Consider a textual representation of an envelope with some dozens of breakpoints, and maybe it won't be so 'digestible' any more... Guess it depends on whether you're digesting/parsing it by eye or by software or what... What's the plan exactly?

Anyway to answer your question, 5 minutes in google got me these hints:

max and pd and processing and matlab and octave do it...

audacity does it...

http://www.kvraudio.com/forum/printview.php?t=210852&start=0

I'm sure a few more minutes will get you something really useful :)

Share this post


Link to post
Share on other sites

There is also the matter of phase offsets to consider in wave packet composition.

Do you actually want to do this on arbitrary wave forms? Are you specifically interested in speech synthesis? A lot of work has been done on speech. You could have a look at formants.

You could have a look at this page describing resynthesis done on a Kawai K5000 addititive synth. If you can come to terms with the name's approach, I think it provides a reasonable walk through for the topic.

Share this post


Link to post
Share on other sites

Whoa - thanks everyone!  I am clearly out of my league here  ;)

The plan was not necessarily for speech but that was the example that came to mind.  I had remembered hearing about FFT a long time ago and for some random reason I thought to myself (in fluent Dummy-ese) "I wonder if we could put a sound back together?"  Sounds like IFFT would do this.  The idea was to use a subset of sines so as to kind of approximate or degrade the signal in specific ways.  It would be neat if I could say "Hi" and then strip out different pieces so as to modulate the sound in various ways.  Or you could replace a given sine with a saw, etc.  At least it would be different than the good old phaser, flanger, filter sweep, etc.

I will follow up on these links.  Again, thanks - this forum has so many smart people on it.

Share this post


Link to post
Share on other sites

Hee hee - I just had another idea.  Maybe I could extract the FFT information, set up the pitches, volumes and durations in MIDI, and then play back a sample of a word on an MB-SID.  Triangle waves would have to suffice.

Might have to wait until I finish the MB-6582 to get more oscillators.  Anyway, f I ever get around to this, the first word will be "Thorsten" and the second "Wilba" ;)

Share this post


Link to post
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!


Register a new account

Sign in

Already have an account? Sign in here.


Sign In Now
Sign in to follow this  
Followers 0