Tau editor and artefacts

Question

Hello everyone,

I’m currently trying to use the Tau editor to morph 2 voices together,

My issue arises when the vocal actors are speaking really quietly. As soon as I convert the wav files into PSI files, everything start to sound weird and 'crackled'. When I solo the 'voiced part only' there's little to no sounds, almost everything as been processed as 'unvoiced part only' even the tonal sounds !
I suspect the Tau to be really confused by the very 'whispery' quality of the vocals, any tips or suggestions ?

thanks in advance!

Comments

When you look at the pitch envelope, does it seem to be correctly tracking the pitch? You might have to re-analyze with a narrow range in order to capture the pitch of "raspy" or "airy" speech.
Is the speech "natural" or has it been pre-processed through a compressor or other effects? If you can get the original unprocessed recordings, it might be easier to get a good analysis.
Could you upload an example?

---

Thanks for your reply, i actually tried both, processed and unprocessed and i'm running into the same issues, the pitch envelope seems to be ok, after reading your comment i've been playing with the range but not getting anywhere, it only gets better as i set the unvoiced threshold to something like 6 or 9 khz  


ps: i'm not sure how to share my psi file in this comment

---

Have you tried using Tools>Spectral analysis on these files to compare the results? (You may have to compress the .psi file before uploading it, or you could send it by email instead).

---

I just generated a classic one as well as a quasi-harmonic analysis, first one sounds a bit blurry and the second one is really metallic/unusable,
overall i think that the Tau one still more 'natural' sounding.

Which email should i send my files to ?
the zip file will be containing the original WAV file,
the psi and spc ones as well, thanks again for the support !