Popping and clicking- interesting observation

[Le DJ Abstrait]

hey, it's me again...

I consider my Wren setup pretty smooth running, but there's one thing that's been bothering me quite a bit: I've been having a rather high rate of clicking/popping. It is rhytmic, sort of related to the patch's cycle, so I ruled out it would be a buffer thing. My audio setup is windows direct sound. I had set the buffer size to zero because I read a few times that with that setting, buffer size is set "automatically".
Well, I just set the buffer size to 2222: clickin's gone!!
Any clues?

[blue hell]

Hi again

Clues .. I can say a little something about it though

Apparently the automatic buffer size suggested by your audio driver is not big enough for slow large patches in Wren.

At the top of the screen there is an indication about what percentage of CPU is being used - when that goes to 60 or higher there may be audio problems .. or alternatively when the fan goes very active there may be audio problems.

My patches tend to run against the limits - on warm days the fan will be noisy - so any computer or audio setup slower than what I have may run out of resources and cause problems.

Using a different driver may help, like ASIO does for me .. but your mileage may vary.

Turning the lights off (ctrl+L) or turning the wires off (ctrl+W) may help a little bit too; setting the sample rate lower may help as well; a larger buffer can help too - but a buffer too large may slow things down again.

I would btw try buffer sizes that are a power of two [64, 128, 256 .. etc.], this does not seem to matter for some audio drivers, but for other's it does. Anyway 2048 may be better than 2222, but only trying and testing will tell.

[I currently run ASIO on an RME interface with a 1024 samples sized buffer @ 48kHz with a control decimation of 32 and a lights rate of 42.24 Hz - but wires are always off and lights are often off too; my computer is an old 15 inch gaming laptop with some I7 processor, the GPU is not being used - and all patches that I publish will run on that setup].

I've tried Wren on some DELL ryzen laptop too BTW, using it's internal audio without setting anything special .. it was almost twice as fast .. that was about ... what is it ... 2 or 3 years ago or so?

[Le DJ Abstrait]

Well, this situation seems a little odd, and that's why I'm writing about it: the patch I'm running is 2019-09-06-02-02 (one of my favorites!). It's not stressing the computer, running at about 12% average according to Wren. What I hear is sounding like there was some DC switching in the audio, which leads me to think it would originate in the patch itself, more than a buffer thing. But then why would it change when i change the buffer size derails this theory...
Bref, for now I'm completely puzzled, trying to understand the situation. I don't think we should lose sleep over this, but it will be a big kick if I can understand the why of the pourquoi...

Yesterday I had friends over here, and I arranged a little electroacoustic parcours with three computers in different rooms running Wren patches, and one running Noodle Radio. We had a long discussion afterwards about alternative realities in music, to the sound of this organic environment...

[blue hell]

[Le DJ Abstrait]

SoCool!!! an electroacoustic jukebox!

It has some similarity with my DJ Abstrait setup, where I have several laptops to play Wren noodles, NMG2 noodles, soundfields and audio extracts, Reaktor ensembles, etc. My setup is less portable, however.

[Le DJ Abstrait]

When the buffer=0, the load is more steady, around 12%
when buffer=2222, there is more variation, like between 9 and 14%.
the relevance of this info is anybody's guess...
I also discovered how to have ASIO working, at last: there must be at least one checkbox enabled in the FW-1884's control panel, for "continuous streaming".
No clicking there either.

[blue hell]

[Le DJ Abstrait]

the computer is an intel motherboard with an i7 quad-core processor at 3.4 GHz. I don't choose my computers, it's another hand-me down from my entourage. it doesn't matter much anyways, the drivers problem seems to be resolved and the sound is good. My next challenge will be to have eSpeak working... even though I have installed eSpeak, I don't hear voices from the patches which employ it. Anything I should know?

[blue hell]

[Le DJ Abstrait]

ah yes, the log! should have thought of that...
here's an extract that that covers most speech-related stuff

0000:01 153 [STARTUP ] probing FFTW DLL presence, looking for "libfftw3-64.dll"
0000:01 198 [STARTUP ] FFTW DLL found (it has no version information)
0000:01 198 [STARTUP ] Looks like an ESpeak installation was found in the registry
0000:01 198 [STARTUP ] probing eSpeak DLL presence, looking for "espeak-64.dll" (using API revision 9)
0000:01 199 [STARTUP ] eSpeak DLL found, version 'ESpeak not initialized'
0000:01 199 [STARTUP ] reading Talkie phrase sets from 'words.lpc'
0000:01 239 [STARTUP ] phrase sets loaded
0000:01 239 [STARTUP ] reading Formant sets sets from 'wren.formants'
0000:01 240 [STARTUP ] formants rules were checked, now reading modal sets from 'wren.modal'
0000:01 241 [STARTUP ] modal rules were checked, some fiddling ...
0000:01 241 [STARTUP ] fiddling done, creating SAPI speech voice names
0000:01 269 [STARTUP ] created SAPI speech voice names

[blue hell]

Hmm

[blue hell]

Hmm eSpeak does not seenm to currently work here either - I can not select any voices.

it just shows 'dyn. id.' meaning that dynamic population of the voicelist did not work.

Weird that, as i did not do any updates whatsover.

Anyway, my log is just as helpful as yours .. soo .. no idea . might look later though ...

[Le DJ Abstrait]

That sucks...

[blue hell]

And in the debug version of wren .. it works ... interresting

I will have to spend a little more time

edit: Hmm .. and when it does work in the debugger .. the log also says that it is not initialized ...

[blue hell]

Right, and I've just recompiled Wren .. and now it works ..beats me

I'll make a new release soon ... with just eSpeak working again if it chooses to do so

[blue hell]

Edit: removed this version - it was not ok yet.

Ok ..made a 2025.4.8.0 version for testing ... it works here now.

I suggest to rename the old wren.exe and then unzip the attached zip and use that as the new wren.exe

I'm running it myself now, it seems ok .. but will make an official release later .. if and when I can remember how to do that again .. it had been a while :lol:

And thank you Le DJ Abstrait for bringing this to my attention :)

[Le DJ Abstrait]

Duh, it don't work!
it starts, and in the green phase of the splash screen it dies.
here is the log it produces.

---

wren.log
Description:		Download (listen)
Filename:	wren.log
Filesize:	2.98 KB
Downloaded:	6 Time(s)

---

[Le DJ Abstrait]

I have removed the Xaudio 2_8.dll file from my Wren folder (which is, if you remember, the illegitimate child of Xaudio 2_9redist.dll that I renamed). Now Wren completes its launch procedure, but complains that Xaudio2_8.dll is absent and just crashes there. I have to cheat about Xaudio2_8.dll at some point, because Xaudio2_8.dll is for windows 8 and I run windows 7...

[blue hell]

https://github.com/rhdunn/espeak/issues/98 .. suggests there to be a solution .. the problem being part of eSpek .. erm .. it looks like it is there ...

[Like ther being a version of eSpeak not needing XAudio at all]

edit: And this one : https://github.com/espeak-ng/espeak-ng/issues/209 - suggesting the same.

[Le DJ Abstrait]

Huh??
Am I supposed to make sense of all this programmer gibberish and the 300 unclear workarounds to this problem??

I'll just go back to my old version, thank you...

[blue hell]

Didn't know you'd go angry- sorry

[Le DJ Abstrait]

I apologise if I seemed rude, but this is clearly not a level of fumbling I'm willing to get into. désolé...

[blue hell]

I had to make some more changes as wren was using a mix of the old eSpeakand the later eSpeak NG; it now strictly uses eSpeak NG.

Some more info is at https://bluehell.nl/blogje/