Wavetables.....Again..... How do I playback only 1 cycle of a wavetable?


#21

The exported .wav is 32 bit float 44.1 khz..... and it is 2048 samples long per cycle.

When I convert it to 16 bit 48khz it becomes 2230 samples long per cycle....

a 128 wavetable is 262144 samples long(32b, 44.1)....

Soooo... They actually fit in Axoloti table sample size without any resizing... But problem is they are 44.1 khz.... I am not sure what is most important sampe size or sample rate? Maybe you know this @thetechnobear ?

I ONLY use Audacity for stretching the samples AND for converting to raw files. All processing/editing I do in Sound Forge Mac version.

Stretching in Audacity:
Use the effect called"change speed". You can change the speed of the file with sample precision. You just put the sample length that you want the sample to be and process it. EASY :wink:

You can do that easily in Audacity with the "change speed" plug in :wink:

Ps... When I am 100% sure everything works in the right way, I will post a full tutorial on how to do the whole process chain, so everyone can make their own sounds for that specifik patch..... I havent tried the version that was posted here yet , cause have been working all day and and today my kids come too... Will check tonight when they are put to bed...


#22

Convert only the sample type to 16 bit, leave the samplerate untouched! Maintaining the same number of samples is important.


#23

Yeah you mentioned that before.. Was wondering what the reason for that is? Axoloti is 16b 48khz, right? The wavetables I made in Serum I sample rate converted to 16 48000 and stretched it to fit table/alloc...... Isnt it better that BOTH sample rate and sample length are the right ones?


#24

It is a waveform defined by 2048 samples per cycle.

Samplerate only becomes relevant when playing it at a certain pitch.

The audio chip is configured tr transmit/receive data at 48kHz/24bit. Patches and objects can use different bit-depths. A normal red or blue cable uses 28 bits nominally.


#25

Ahh... So if you use 44.1 you wont get right pitch on each note when played on a keyboard? Then I would say that it is not good to advice people to use 44.1.. Better to stretch it and use right sample rate, right? It is important that the pitch is right. At least that is my thoughts.

But like I mentioned, I'll make a tutorial so everyone can enjoy wavetable in an easy way. Ity is actually really easy to make them from Serum. But I just want to make sure I got all the data right before I start making that tutorial.


#26

a single cycle is independent of SR/pitch... its the 'speed' you play it back that gives you pitch.
(remember : frequency = Hz = cycles per second)
its the cycle thats gives you pitch... this is why samples, which are multi cycle ARE pitched.

thanks for all the info... Im quite liking serum, its a good way to create wavetables, and a nice synth in its own right.
... im also wondering if I should given audioterm another chance, perhaps playing with it a bit, It might 'click'

question - how many samples to use for a wave? serums 2048 seems quite high, which is fine for a PC, but I'm wondering given the limited memory on axoloti, what could we go down to? 1024/512/256 ?


#27

I only have a PC, one of those notebooks, which are limited.. And it cant run the Audio term editor, so I never tried it...

But I used this one alot for making wavetables for Blofeld. It exports to both Blofeld specifik format and wavefiles(Pc only, though):
http://lady.rdsor.ro/~kotro/index.php?m=soft#mopho


#28

I am pretty sure that there is a difference if you use 44.1 and 48.. Will test how a wavetable exported from Serum without any editing and a version that is converted to 48khz ans stretched to fil table sounds compare to each other.

Blofeld uses 64X128=8192 samples for a whole wavetable.

I think Serum wavetable size depend on how big you make it. Most wavetables is fixed 128 waves, but in Serum you can make up top 256 waves..... But that doesnt make any sense in relation to Axoloti which has 128 steps knobs. So I only make 128 waves and they are ALWAYS same size:

2048X128=262144 samples

One way to do it:
You could try stretching them to half size with the tempo change effect or maybe even quarter size? That shouldnt matter, right? It is the same as pitching the sound up and octave or two up....

Before I tried the change speed effect I was sceptic about stretching them, but for small amounts it seems to work well. Havent tried larger stretches, though.

EDIT:
BUT.. If the sample rate really doesnt matter a Serum wavetable with 128 waves ONLY converted to .raw should fit table/alloc with 262144 sample size perfectly. If that is the case, it is really awesome.... Anyway still need to test the example posted earlier in this post


#29

yeah, I think SR is just required as the wav file format requires it, so if you playback a wav file it knows what the SR is (amongst other data)... but we are using RAW so we strip all that header information off.

whats interesting in Audacity (and DSP-quattro) you can change the number of samples used either by changing speed, or the sample rate... and I think its an identical operation IF you then save as RAW.

been playing with audioterm a bit, looks comprehensive, and reasonably quick to get around BUT I thikn I'm going to need to find a video/document on how to achieve some simple functions (e.g. resampling and to see what things covers, and what it doesnt... but it is massive, not only wave editing, wave tables but also slicing and who knows what else :smile:

I think Id really like to hear some 128/256 sample waves, these could really convert me...


#30

They might be good, havent tried that short ones. But Ill make one in Serum and then stretch it to 256 and check if it works... Would be nice not to use so much SDram on an oscillator.


#31

yeah, exactly I'm thinking the same... in particular you can get some small single cycle waves of old analog gear, which I think could sound quite nice, at the right resolution. ... and of course with wavetables gets even more 'critical'


#32

Serum also has an export format called "8-bit .256). It is for using with older wavetables synths and can be burned directly to a ROM. Wonder if this format can be used for anything?


#33

@keyman

Just made a patch from the picture you posted and it sounds pretty close if not identical to the same wavetable loaded in Serum.... will need to do some more testing on that. But very positive about the result :smile: Thanks to everyone :smile:

@thetechnobear

Here is a few different versions of the same wavetable I posted earlier. If we use that one, then we have something to compare with. They are all stretched in Audacity. The names are:
128NewM1.raw=1024 samples
128NewM2.raw=2048 samples
128NewM4.raw=4096 samples
128NewM5.raw=512 samples
128newM6.raw = 256samples

I just tested one of them, and to me it didnt sound right. Check for yourself.
Link:
https://mega.nz/#!TVo32ILY!4qLxezxYcspX7_iKdktmJ_4ThzAtSWCbFLUPQL3OTC8


#34

I'm here (night session :wink:
I can post the patch…Wavetable 00.axp (5.0 KB)
Will download and read up..


#35

Adjusted number of samples accordingly for each raw file.

128newM1 - this is 0.00 (first cycle)

5.00 so its 11th cycle (octave below before 12th)

128newM2 - this is 0.00 (first cycle)

5.00 so its 11th cycle

128NewM4 - this is 0.00 (first cycle)

5.00 so its 11th cycle

128NewM5 - this is 0.00 (first cycle)

5.00 so its 11th cycle

128newM6 - this is 0.00 (first cycle)

5.00 so its 11th cycle


#36

Had to go to bed last night so didnt have time to look. I see now they look very different.... And they also sound a lot different from what I heard last night.

@keyman what do you think?

Edit


#37

@jaffasplaffa yes all versions sound different as scope shows also; not sure what to look for? many "variables" if you change table size, mix/mix 1 amount…
I have DL some free wavetables here and there will try with Alchemy compare some and convert some others, but point is we can "scroll" between waves :wink: right !


#38

I think in perfect world they should look identical, but some of them are stretched around 1000%..... I think stretching a little bit is ok, but 500%-1000% will definatly affect the sound in one way or the other.

Yes, exactly. Thank you for the input :wink: The waves are indexed properly for 128 waves wavetable as long as the sample is exactly same samplesize as the table/alloc.


#39

I stand corrected :smile: I just had to test it myself. Havent worked with such small waves befor and was not sure about the behaviour.

Anyway, I did a test and came to the result that a wavetable from Serum with 128 waves sounds EXACTLY the same when loaded in Axoloti in these two scenarios:
1. Exported from Serum and converted to 16-bit and then to.raw = 262144 samples
2. Exported from Serum converted to 16-bit 48khz, stretched to 262144 samples and then to .raw.

Thanks for pointing it out. That REALLY makes it easy to make wavetables Or use from web....... most of them are 128 step and you only have to adjust table/alloc size to same size as the length of the wavetable in samples. Awesome :smile:


#40

Was wondering how to change the number of steps of the dial on Axoloti from 128 to 64 to fit wavetables with 64 steps? I tried changing the dial p to a dial b, and use only the positive vaules and that works, but would be really nice to use the dial p instead. I tried with a div2 after the dial p but that didnt seem to work.

Any ideas?