Hi, I have been playing around with resynthesis. With that in mind when I import a sample I expect it to sound nearly identical to source material. When I imported a generic monotone vocal sample I found some issues. I believe I imported everything correctly. I watched a ton of tutorial videos and examined the manual. However every time I import a vocal sample it doesn't sound very good. I have to play very low notes c1, c0…to get the formant content. Also the waveform at this low pitch cycles very slowly so the end result sounds like a person talking through a fan. Kinda like Chris Farley in Tommy boy.. My concern is that if resampling is inaccurate for vocal samples it's inaccurate for all samples of moderate complexity. How right or wrong am I and what can I do to improve resampling quality…..
Serum isn't a sampler; so unless the waveform your importing is only one cycle of the wave, you won't audibly get the same result as what you're importing. What you're likely hearing is one little wave cycle of a sample that is comprised of thousands of cycles changing over the course of the clip.
If you want your vocal sample to re-pitch, then use a sampler ;)
The resynthesis of a wave is great for generating digital waveforms that closely emulate analog waves. I've gotten some great textures on waves that I've recorded on a moog and then resynthesized into serum. It allows me to use these analog-like waves as oscillators in FM and RM modulation configurations within serum. It's just a way to extend your pallet of waveforms outside what comes built into serum.
I understand the difference between sampling and resynthesizing. Also before I continue its not a matter of life or death issue, its just something about serum I found interesting and that I want to explore further….So I actually recorded an example. This is a minute long track that consists of wavetable resynthesized vocals. The first one is using dune and an external wavetable tool called audio term. it creates wavetables from samples and does a good job. You can hear that it sounds synthesized, it is also repitched properly and not sped up like a sampler would do, and inteligible across the pitch range…
https://soundcloud.com/loris-solic/synthesizer
The second vocal saying "round one" is created with serum. I tried playing it at the same pitch as the other sample like c2 to c3 range but it sounds like screaming death. so I played some lower notes c0 c1. this is the result. sounds good..but you can hear that its not as smooth and unresponsive to pitch changes. I feel that serum should be capable of creating the same quality wavetables from samples as audio term and I just dont know how to do it….
If you use an envelope to modulate through the wavetable of the clip you're resynthesizing from, you can get a result similar to the one you got using Dune / audio term.
Here's a recording of a word being re-pitched quite cleanly via serum. The original clip being sampled / resynthesized plays at both the beginnning + end:
Very nice, I know how to use the envelope to cycle through.. tell me the process you used to import that waveform, under which condiitons did you import.
I tried importing at 256 and 512 samples and it seems to be working better….
I'll add that in your soundcloud example, notice that "synthesizer" is monotone (no pitch change) where as your second example has a ton of pitch bend. The best solution really is to use a source with a perfectly flat pitch (such as processing the speech first using a vocoder with a single note as carrier) or perhaps (probably slightly less ideal) removing all pitch with auto-tune/melodyne.
Then you can import the audio using a constant framesize as the import option.
hope this helps,
Steve
I imported using constant frame size; however, you can sometimes get better results using the other options, based on the pitch content of the clip you're importing.
Also, like Steve said, if you tune the clip first, you'll get a much more clean result. I used Waves tune on the audio clip in the soundcloud link i posted before resynthesizing it into serum.
Btw have u ever used vocal pack from here?
https://www.lucidsamples.com/42-acapella-vocals-samples
I still thinking about buying cuz i rly looking for
some high quality acapella.
