It basically is a banality - but is the sample causing all this is in the right format? And stripped of all metadata? Sometimes, when including lots of external stuff, I tend to overlook details like this. Could happen to everybody I reckon?
Did you try and upload the Set WITHOUT the sample in it at all. That’s really important for us trying to debug.
NOTE: When the Creator is online… you’ll get the error directly on upload of the individual sample… and not right at the very end, which will be good!
From MediaInfo:
General
CompleteName : D:\Content\Audio\D&D\Beyond Death\Ambiance\Inferno_14595_1460041913 - Copy.ogg
Format : Ogg
FileSize/String : 2.36 MiB
Duration/String : 36s 728ms
OverallBitRate_Mode/String : Variable
OverallBitRate/String : 540 Kbps
Encoded_Application/String : libsndfile
Audio
ID/String : 571981636 (0x2217BF44)
Format : Vorbis
Format_Settings_Floor : 1
Duration/String : 36s 728ms
BitRate_Mode/String : Variable
BitRate/String : 500 Kbps
Channel(s)/String : 2 channels
SamplingRate/String : 44.1 KHz
Compression_Mode/String : Lossy
StreamSize/String : 2.19 MiB (93%)
Encoded_Library/String : libVorbis (⛄⛄⛄⛄) (20150105 (⛄⛄⛄⛄))
Should probably be 1 channel for my use, but still… that’s the file Audition spit out.
Without that element (or it’s duplicate) at all it still doesn’t work. For the sake of completeness here is an updated JSON. I think it’s the same just smaller… but I may have added more elements or something. Hard to remember clearly at this point.
And I’m already sold on the move to Syrinscape Online 
. Definitely looking forward to that.
Honestly, at this point, it seems like I’m just going to have to use the creator UI and hope for the best. It is what it is.
I would still like to know what is going on though so I can avoid it happening again.
ARG! I just found that my samples had lost their copyright text again!
EDIT: and they lost their license text yet again! This time all I did was reload the creator ui. :
:
EDIT 2: Went one by one through every sample in the set setting and resetting each license text and reloading the ui till it “stuck”. Then uploaded the set again… didn’t fix the problem.
How many “external” samples did you use in the soundset(s) that don’t work? Yes, the creator is a bit “picky” from time to time and it can happen really fast that you overlook a thing or two while creating, especially if you’re in the workflow and work “too fast”. My bet is on a faulty sample somewhere…
30 imported samples. But about 16 of those are also used in other soundsets that work fine. Also, a many of my elements use 3-6 of these samples… so the number of elements with imported samples is much lower. All samples work in the creator… in fact I ended up using the creator in my session tonight (didn’t have an alternative)… in the creator everything worked flawlessly (which is actually unusual).
I use a template in Audition for exporting the samples to ogg… so every single file has the exact same stream and container formatting. That format is identical to the samples stored in the media/sample folder of Syrinscape (at least according to MediaInfo). Metadata is always stripped. The files still get transcoded when importing into Syrinscape Creator though.
Of course even with all that, it’s still possible I missed something.
OK!
@JadedDragoon Since this hasn’t got fixed in any of the normal ways.
Can you actuall export the entire upload zip and share the whole thing with me, and I will try and upload it myself.
Note: It’ll need to be less then 100MB.
THEN we can REALLY sort this out once and for all! 
Here it is: (@steve removed link and sent via emai to @benjamin)
@new_vision Isn’t wrong that the problem is probably something stupid simple. But I’m just not seeing it. Really would love to know what went wrong.
P.S. This link is not intended for distribution. I make no claim of ownership of the content within. Any download of the above soundset for any reason other than diagnosing this issue and uploading to the syrinscape servers is the responsibility of the downloader. The above soundset is expressly not intended to be shared for the use of anyone other than myself and those uses listed above.
Failed. Boo! 
Something wrong with this sample.
“Inferno_14595_1460041913 - Copy.transcoded.ogg”
Next steps:
Same sample as before.
Have you tried copying that sample into a completely fresh wav. 44.1K 16 bit Stereo, No tags. Yes?
Send me that sample.
See this is getting really weird at this point.
There are actually two instances of that sample in this soundset. The original source files are:
- Inferno_14595_1460041913.ogg
- Inferno_14595_1460041913 - Copy.ogg
The copy was made with a file copy operation before both were imported into syrinscape. As such they should be digitally identical to each other. But just to be sure, I checksumed them:
-
Inferno_14595_1460041913.ogg
- 3cc3fffd23c356e42076c5c996ffc26a5196abe614ff334a6a4ca42997a329f2
.
- 3cc3fffd23c356e42076c5c996ffc26a5196abe614ff334a6a4ca42997a329f2
-
Inferno_14595_1460041913 - Copy.ogg
- 3cc3fffd23c356e42076c5c996ffc26a5196abe614ff334a6a4ca42997a329f2
Identical checksums, right? Should be identical files then… right? So why is there no error on the first file?
For completeness, mediainfo output of the source sample:
General
Complete name : D:\Content\Audio\D&D\Beyond Death\Ambiance\Inferno_14595_1460041913.ogg
Format : Ogg
File size : 2.36 MiB
Duration : 36s 728ms
Overall bit rate mode : Variable
Overall bit rate : 540 Kbps
Writing application : libsndfile
Audio
ID : 571981636 (0x2217BF44)
Format : Vorbis
Format settings, Floor : 1
Duration : 36s 728ms
Bit rate mode : Variable
Bit rate : 500 Kbps
Channel(s) : 2 channels
Sampling rate : 44.1 KHz
Compression mode : Lossy
Stream size : 2.19 MiB (93%)
Writing library : libVorbis (⛄⛄⛄⛄) (20150105 (⛄⛄⛄⛄))
So… digging deeper still I dug up the transcoded copies of the files from syrinscapes sample folder and check-summed those (BTW, I’d really like more control over where syrinscape stores its files… it’s eating up all the space on my 128 GB SSD while barely utilizing my 6000+ GB RAID10 array at all):
-
Inferno_14595_1460041913.transcoded.ogg
- cbea424a1e42820f07fb8a753a71c653d477a4eb869c4a200c42fd12505ff0e1
.
- cbea424a1e42820f07fb8a753a71c653d477a4eb869c4a200c42fd12505ff0e1
-
Inferno_14595_1460041913 - Copy.transcoded.ogg
- 958e262cb83e8f253ae374ff3afd83a69378f5ffe60ed2e28495a146e1dd759a
Huh… that’s interesting. Give the same input to Syrinscape twice in a row… get two different outputs. So I looked at syrinscape’s sample files with mediainfo:
General
Complete name : D:\Content\Audio\D&D\Beyond Death\Ambiance\Inferno_14595_1460041913.transcoded.ogg
Format : Ogg
File size : 614 KiB
Duration : 37s 728ms
Overall bit rate mode : Variable
Overall bit rate : 133 Kbps
Writing application : libsndfile
Audio
ID : 1579170999 (0x5E203CB7)
Format : Vorbis
Format settings, Floor : 1
Duration : 37s 728ms
Bit rate mode : Variable
Channel(s) : 2 channels
Sampling rate : 55.0 KHz
Compression mode : Lossy
Writing library : libVorbis (Turpakäräjiin) (20140122 (Turpakäräjiin))
and
General
Complete name : D:\Content\Audio\D&D\Beyond Death\Ambiance\Inferno_14595_1460041913 - Copy.transcoded.ogg
Format : Ogg
File size : 614 KiB
Duration : 37s 728ms
Overall bit rate mode : Variable
Overall bit rate : 133 Kbps
Writing application : libsndfile
Audio
ID : 60963579 (0x3A23AFB)
Format : Vorbis
Format settings, Floor : 1
Duration : 37s 728ms
Bit rate mode : Variable
Channel(s) : 2 channels
Sampling rate : 55.0 KHz
Compression mode : Lossy
Writing library : libVorbis (Turpakäräjiin) (20140122 (Turpakäräjiin))
Ok then. The only difference in formatting is the audio streams have different stream ID’s. I’m chalking that up to an oddity of the ogg container format. But it would be enough to change the checksum.
It would not, however, be enough to allow one transcoded sample to work and the other to fail. Given there is no difference in container and stream format beyond the stream IDs… I have to assume there is an error in the stream itself.
What makes an inconsistent transcode hard to swallow is the fact that I’ve deleted and re-imported these samples many times. I repeat, both samples have been deleted and trans-coded from the source samples multiple times. And each time “Inferno_14595_1460041913.ogg” works and “Inferno_14595_1460041913 - copy.ogg” does not. If the transcoding process was sometimes resulting in corrupt audio streams and other times not while being fed the same input in all cases… one would expect the malfunction to be somewhat random… not extremely consistent. Also… the literal only difference between source samples is the filename. How would the filename be causing this problem?
All we need now is a late rabbit and a madman in a top hat and it’ll be a tea party.