Hello everybody:
I have a table filled with audio data, for instance, 20s of a singer line. I would like to get the frequencies of that recording for every window of W samples (W=2048, 4096 or so). I can get those single values in time using [sigmund~] but I cannot copy them on the fly into a table. That vector (whose length is the same length as the audio) would have constant values for each window length.
The idea, later on, is to have an out of tune detector.
Thanks in advance.

Sumidero

[tabsend~] / [tabreceive~]

does that help?

Couldn't you do something like this?

[sigmund~]
|
[t f b]
| \
| [f 0]X[+ 1]
| \
| \
[tabwrite pitch]

Hello Maelstorm, mod:
It seems that I have to take a different approach. [ tabsend~ ] / [ tabreceive~ ] only wrote the first <blocksize> elements of the table. I have to work around my entire patch to implement something using either [ tabwrite ] (for a stepped pitch table) or [ poke~ ] (for a continuous variation of pitches). Let's do some vector math-surfing!
Thank you very much, indeed. I learnt something from this.

sumidero

If you want something continuous, why not run [sigmund~] through [vline~] and write it with [tabwrite~]? Sorry if I'm not fully understanding what you need.

Hello Maelstorm:
You do! I tried your last suggestion before but seems that I failed in doing it right. I have the patch at home so I can't attach it now, but my first idea was to pack a couple of floats per sigmund~ output (destination pitch and time window elapse) and send them via message to a vline~ into tabwrite~, but it seems that I wired something wrong and it did't work.
I'm working on a different approach using a table with a single pitch per analysis window and using tabread4~ to do the interpolation. Unfortunately I have tons of other work to do so I can't dedicate the time I'd like.
Thank you all very much. As soon as I solve this I'll post a solution.

Sumidero

Hello again:
After some time I found out that granular synthesis was not the optimal choice for what I wanted to do.
Here is the patch that works good with simple tunes but does not work properly with recorded voices. I built it based on I07-phase-vocoder.pd example file. What I meant to do is something like what the commercial software Melodyne does.
I tried what you said , Maelstorm, about storing pitches in a table, and it worked (thanks a lot, of course) Do I need some more complicated tune detector? Is is some smoothing issue?
Any help will be very appreciated.
Best regards,

Sumidero

http://www.pdpatchrepo.info/hurleur/autoTunerPlayer03.pd