[Csnd] GEN52 issue

The Csound Mailinglist
1

Hey hey,
I try to combine two audio channels from an audiofile into one interleaved table using GEN52:
iLeft = ftgen(0, 0, 0, 1, "input.wav", 0, 0, 1)
iRight = ftgen(0, 0, 0, 1, "input.wav", 0, 0, 2)
iLength = nsamp(iLeft)
iInterleaved = ftgen(0, 0, (iLength * 2), 52, 2, iLeft, 0, 1, iRight, 0, 1)

This fails with:
INIT ERROR in instr 2 line 57: deferred-size ftable102.000000 illegal here
iIR ftgentmp.i 0 0 #i13 52 2 iLeft 0 1 iRight 0 1
This fails irrespectively of a negative or positive table size, or whether the two channels come from one multichannel GEN01 table or several single channel GEN01 tables.

The task is: to get a stereo interleaved table for ftconv, never mind how many channels it has, as long as it's >=2 channels.

Tested with one stereo file directly passed into a GEN01 and then used in ftconv it works. So the file and table read operations up to that point work.

Any ideas or tips to further tackle the problem would be very welcome!

Best wishes,

Jeanette

2

sounds like a bug to me, and i think you should open an issue for it.

a workaround might be to write the first two channels of your input soundfile with a UDO to a stereo file, if this file has more channels than 2, and then use it in the way your describe.

another workaround could be to get the required length of the GEN01 table from the soundfile, and then pass it to ftgen as argument. so something like
iTableLength = filelen("input.wav") * filesr("input.wav")
iLeft = ftgen(0, 0, iTableLength, 1, "input.wav", 0, 0, 1)

if this works, it is certainly better than writing a file to disk ...

3

Hi Joachim,
thanks for the workaround tip.s The second one is good. I may use this. For the time being I have limited the code to only accept stereo files, but I'd like to lift that restriction.

Best wishes,

Jeanette