Tech Preview: Hive Wavetables

[Urs]

I never really got hooked on the sound of wavetable based synths mostly because of the audible artifacts. But some of the posted examples here sounds incredible.

But since I already own Hive I really want to dig into making my own wavetables for it. And some of the examples posted here sounds really inspiring.

I got a pretty large eurorack with some incredible sounding oscillators, filters and wavefolders etc. It would be so cool to create some tones and import these into Hive.

I read the entire thread but I can't find any way on how would go about creating my own wavetables based on my own samples and importing them into Hive?

That's the problem... creating wavetables from sampled sources is exactly what's usually causing some of these artifacts. I've always been opposed to Wavetable synthesis for this very reason.

Visual editors can make artifact-free wavetables, but it's tedious. This is why we added the scripting engine. It seems tedious at first glance, but in my opinion it's faster to get great tones than using a visual editor. And you can do things you wish you could do in visual editors, such as formant synthesis.

- U

[Delta Sign]

Yeah, I pretty much deleted all my sampled wavetables after I started messing with the scripting

Correctly sampling wavetables without any artifacts is hard. Like, I even made a PD patch that generates and samples wavetables at exactly 23.44hz @ 48khz (that's exectly 2048 samples per single cycle) and does the changes over exactly the right amount of time to get 256 frames, but even that is just tedious and not without problems. Sampling a synth that maybe even has drift, noise etc. would be kinda impossible.

The "sampling" in Zebra works really well, though. I still like to get single cycles from that, even if I mess with them with the scripting later.

[fmr]

Every time I change the interpolator to Spectral Fade, Hive crashes.

[fmr]

Another thing, again regarding WAV wavetables (sorry to bother with this).

I tested Reaktor wavetables, and they translate well.

However. I tried also PPG wavetables, and they do not translate. I was checking the files, and they seem to have a much smaller duration. While Reaktor files have a duration of 1.533 seconds, the PPG files have a duration of just 372 ms. This seems correct to me, because the PPG wavetables have just 64 waves, while the Reaktor ones have 256 waves (4x). This should bve solved by adding "-64" to the PPG wavetables. However, even adding that doesn't make Hive to "read" the wavteables as it should - it basically reads eight waves as one cycle.

I tried to pitch down the WAV file to change the durewtion. This solves the problem, but the waves become distorted (no longer the perfect waves I have in the original WAV file)

Any hint or tip?

[exmatproton]

Another thing, again regarding WAV wavetables (sorry to bother with this).

I tested Reaktor wavetables, and they translate well.

However. I tried also PPG wavetables, and they do not translate. I was checking the files, and they seem to have a much smaller duration. While Reaktor files have a duration of 1.533 seconds, the PPG files have a duration of just 372 ms. This seems correct to me, because the PPG wavetables have just 64 waves, while the Reaktor ones have 256 waves (4x). This should bve solved by adding "-64" to the PPG wavetables. However, even adding that doesn't make Hive to "read" the wavteables as it should - it basically reads eight waves as one cycle.

I tried to pitch down the WAV file to change the durewtion. This solves the problem, but the waves become distorted (no longer the perfect waves I have in the original WAV file)

Any hint or tip?

Maybe open them in Serum and use any morph tool to extend them?
If you don't have Serum please post one PPG table so i can test it

[PietW.]

Here is a Wavetable create with Audioterm. Exactly 2048 Samples and 256 per Frame.
I think here are no Artefacts or Crackles.

https://www.dropbox.com/s/kms3bjyg0fsyh ... IE.7z?dl=0

[fmr]

Another thing, again regarding WAV wavetables (sorry to bother with this).

I tested Reaktor wavetables, and they translate well.

However. I tried also PPG wavetables, and they do not translate. I was checking the files, and they seem to have a much smaller duration. While Reaktor files have a duration of 1.533 seconds, the PPG files have a duration of just 372 ms. This seems correct to me, because the PPG wavetables have just 64 waves, while the Reaktor ones have 256 waves (4x). This should bve solved by adding "-64" to the PPG wavetables. However, even adding that doesn't make Hive to "read" the wavteables as it should - it basically reads eight waves as one cycle.

I tried to pitch down the WAV file to change the durewtion. This solves the problem, but the waves become distorted (no longer the perfect waves I have in the original WAV file)

Any hint or tip?

Maybe open them in Serum and use any morph tool to extend them?
If you don't have Serum please post one PPG table so i can test it

OK, here is a link to the wavetable #31:

https://www.dropbox.com/s/sr7vbbtv6w64r ... 4.wav?dl=0

I also extracted the individual waves, so, I can recreate the wavetable using a script to point to each of the waves (a convoluted and tedious job, but I think it will work, if someone instructs me how to come up with such a script),

Using Serum will create extra waves, and that's what I wanted to avoid (or am I missing something?)

[fmr]

I just came up with another simple script. It's basically a sine wave with gradually lowering bit resolution during the first half of the table. The bit resolution ramps up again in the second half but noise is also gradually introduced.
There are some cool bad radio transmission sounds around the middle of the table

Code: Select all

NumFrames=128
Wave "sin(2*pi*phase)"
Wave "round(((table+0.04)*32)*x)/((table+0.04)*32)"
NumFrames=256
Move start=0 end=127 to=128
Wave start=0 end=127 "main_fi(127-(frame-128), index)"
Wave end=0 "sin(2*pi*phase)"
Move start=0 end=0 to=255
Wave start=128 blend=add "rands*((frame-128)/127)"
Spectrum lowest=0 highest=0 "0"
Normalize base=each

Again, I'm still learning the scripting, so the script is probably totally nonsensical, but it works

I tried this, and it isn't working here (but the previous script you did, 2Op FM worked great - )

[fmr]

Here is a Wavetable create with Audioterm. Exactly 2048 Samples and 256 per Frame.
I think here are no Artefacts or Crackles.

https://www.dropbox.com/s/kms3bjyg0fsyh ... IE.7z?dl=0

Yes, this one scans perfectly

[exmatproton]

Another thing, again regarding WAV wavetables (sorry to bother with this).

I tested Reaktor wavetables, and they translate well.

However. I tried also PPG wavetables, and they do not translate. I was checking the files, and they seem to have a much smaller duration. While Reaktor files have a duration of 1.533 seconds, the PPG files have a duration of just 372 ms. This seems correct to me, because the PPG wavetables have just 64 waves, while the Reaktor ones have 256 waves (4x). This should bve solved by adding "-64" to the PPG wavetables. However, even adding that doesn't make Hive to "read" the wavteables as it should - it basically reads eight waves as one cycle.

I tried to pitch down the WAV file to change the durewtion. This solves the problem, but the waves become distorted (no longer the perfect waves I have in the original WAV file)

Any hint or tip?

Maybe open them in Serum and use any morph tool to extend them?
If you don't have Serum please post one PPG table so i can test it

OK, here is a link to the wavetable #31:

https://www.dropbox.com/s/sr7vbbtv6w64r ... 4.wav?dl=0

I also extracted the individual waves, so, I can recreate the wavetable using a script to point to each of the waves (a convoluted and tedious job, but I think it will work, if someone instructs me how to come up with such a script),

Using Serum will create extra waves, and that's what I wanted to avoid (or am I missing something?)

Well, i was able to extend it using Icarus. Serum gave me poor results. I am not sure if this is what you mean though. It is a faded WT now. It sound good in Hive though, but as a continues WT

https://www.dropbox.com/s/srek4ct2odzt0 ... 1.wav?dl=0

[fmr]

Well, i was able to extend it using Icarus. Serum gave me poor results. I am not sure if this is what you mean though. It is a faded WT now. It sound good in Hive though, but as a continues WT

https://www.dropbox.com/s/srek4ct2odzt0 ... 1.wav?dl=0

It works the same as it was before. If you look at the wavetable window, you clearly see that there are eight cycles in the window (there whould be just one, IMO).

It works (kind of) but not as it was supposed to - you get that "buzz" effect that was mentioned before, because you are reading several cycles as a single one.

I could assemble a wavetable in Audio-Term out of single-cycle waves, but I cannot get Audio-Term to recognize "these" single-cycle waves I extracted

[pdxindy]

I just came up with another simple script. It's basically a sine wave with gradually lowering bit resolution during the first half of the table. The bit resolution ramps up again in the second half but noise is also gradually introduced.
There are some cool bad radio transmission sounds around the middle of the table


Anyway, here is the script in a somewhat granular sounding patch:
https://soundcloud.com/deltasign/hive-wavetable-test-07

Gorgeous!!

[exmatproton]

Well, i was able to extend it using Icarus. Serum gave me poor results. I am not sure if this is what you mean though. It is a faded WT now. It sound good in Hive though, but as a continues WT

https://www.dropbox.com/s/srek4ct2odzt0 ... 1.wav?dl=0

It works the same as it was before. If you look at the wavetable window, you clearly see that there are eight cycles in the window (there whould be just one, IMO).

It works (kind of) but not as it was supposed to - you get that "buzz" effect that was mentioned before, because you are reading several cycles as a single one.

Ah yes. That's the interpolation of Icarus i guess (and in Serum for that matter). Sorry i couldn't be of more assistance

[PietW.]

Here is a Wavetable create with Audioterm. Exactly 2048 Samples and 256 per Frame.
I think here are no Artefacts or Crackles.

https://www.dropbox.com/s/kms3bjyg0fsyh ... IE.7z?dl=0

Yes, this one scans perfectly

Fine. I'm glad.
Thank you.

[fmr]

Here is a Wavetable create with Audioterm. Exactly 2048 Samples and 256 per Frame.
I think here are no Artefacts or Crackles.

https://www.dropbox.com/s/kms3bjyg0fsyh ... IE.7z?dl=0

Yes, this one scans perfectly

Fine. I'm glad.
Thank you.

You're welcome. Can you assist me in getting the single-cycle waves I extracted being recognized in Audio-Term? Seems like AT only recognized the syngle cycle waves it creates itself