Realtime Detune ???

Status
Not open for further replies.

Ticasso

New member
Hi,

Sorry for my bad English :(

I’m learning and using a Teensy with the Audio Shield and I ‘would like to know if it is possible to make a “realtime” detune sound
From the input to the output.

I’ve try the different effects like in the “48 mn” video but I didn’t succeed :(

If someone could help me or telling a way to go, it will be really really clever

Thanks
 
I don't quite understand what you're asking.

Maybe you could explain with an example or a link to software or products that do this? (if any exist)
 
Maybe this algorithm? (which might be too much for Teensy... or maybe not?)

http://www.guitarpitchshifter.com/pitchshifting.html

I think the teensy can do that.
The teensy can decode mp3 and process fft on the result (should be comparable effort (or more since it involves additional huffman-decode + other "details"..))

Here's some code, but it is floating point und must be converted to fixed point and adopted to the library (i'm not sure about the ram usage..):
http://blogs.zynaptiq.com/bernsee/pitch-shifting-using-the-ft/
 
Last edited:
the best way to do this is by modulating the time of a short delay effect and listen to only the wet delay. There is an artifact that happens when you change delay times smoothly that causes the pitch to bend up and down along with the delay time. teensy has a 4 tap delay object, but i am not sure if it has smooth time interpolation needed for this to work. without smoothing on the delay time changes at a per sample level you will end up with an abrasive digital click instead of the pitch warble.

using a pitch shifter effect can potentially work,it depends on the nuance of the type of sound you are looking for. most pitch shifting techniques tend to cause slight digital artifacts.

Another big part of what gives the whammy bar its feel are the performance gestures which are heavily based on the physical act of playing the bar. If you are trying to get a natural sounding whammy you will have to study and possibly capture performance gestures if you can find a good way to do that.
 
Last edited:
Maybe this algorithm? (which might be too much for Teensy... or maybe not?)

http://www.guitarpitchshifter.com/pitchshifting.html


Wow, that would be amazing, a phase vocoder on the teensy. That is one of my favorite algorithms. There is a phase vocoder pure data patch in the built in example library that only takes 2-3% cpu on my laptop, however it does the analysis first and then allows playback of the analysis buffer with transposition and very smooth interpolated time scrubbing through the captured bins. When I was talking about pitch shifting in my previous message I was thinking of the simple grain-type technique of playing back windowed transposed wavelets, which I have tried using the wavetable object. it came out interesting but very digital sounding. That the of thing would for sure run on teensy.

Im reading this article you posted, after seeing this pure data patch is way lower on CPU then I had imagined this seems like a pretty cool project.
 
Hi,

Thanks a lot for your interest :)
The pitch bend effect is really an "old school" effect on a synth.
I think that it will be louder with real sounds. i've haven't find any other Teensy projects about it.
i've also test the Mozzi Audio with an Arduino, but doesn't feet with Real sounds to.

@+

Edit : I've just find some Matlab Algo here : http://www.cs.cf.ac.uk/Dave/CM0268/PDF/10_CM0268_Audio_FX.pdf
They seems to be "guitar" effects that i'm using with my guitar effect hardware racks.
The Vibrato one and wah-wah seem to be quite like i will hope.
i don't know if it can be translate in Teensy ....
 
Last edited:
Status
Not open for further replies.
Back
Top