This browser is no longer supported.
Upgrade to Microsoft Edge to take advantage of the latest features, security updates, and technical support.
![](data:image/svg+xml, %3Csvg xmlns='http://www.w3.org/2000/svg' height='64' class='font-weight-bold' style='font: 600 30.11764705882353px "SegoeUI", Arial' width='64'%3E%3Ccircle fill='hsl(118.4, 50%, 21%)' cx='32' cy='32' r='32' /%3E%3Ctext x='50%25' y='55%25' dominant-baseline='middle' text-anchor='middle' fill='%23FFF' %3EBL%3C/text%3E%3C/svg%3E)
Hi,
We are currently starting a rather large transcribtion project, and we have been training the model quite extensively. We have, however, run into a problem I cannot seem to solve. I have around 100 files (out of 1500ish), which for some reason just refuses to load correctly into the training module. I get the following message on all of them: message": " Error: audio file does not exist
All the files has been formatted exactly the same (plus, I have tried to change the files which gives me errors to a couple of different HZ just to see), and it still returns the same. I have also tried to upload them individually, changed the names around, and re-rendered them all from stratch. There is no common denominator in wording, length or anything.
Is there anything here I am missing? Or what can I do to fix it?
Looking forward to hearing from you!
Best,
Jens
![](data:image/svg+xml, %3Csvg xmlns='http://www.w3.org/2000/svg' height='64' class='font-weight-bold' style='font: 600 30.11764705882353px "SegoeUI", Arial' width='64'%3E%3Ccircle fill='hsl(22.400000000000002, 11%, 12%)' cx='32' cy='32' r='32' /%3E%3Ctext x='50%25' y='55%25' dominant-baseline='middle' text-anchor='middle' fill='%23FFF' %3ERM%3C/text%3E%3C/svg%3E)
@Blueprint Learning Thanks for the question. Can you please share the sample file to repro the same.
Can you please add more details about the document that you are trying, Here is link to custom speech.
Hi,
Certainly! :)
I've linked a wetransfer link (https://we.tl/t-Tk8Me3Vi5g) here, with the exact zipped folder of the structure we are using, with the words which doesn't work. As mentioned, I have already managed to a lot more words to pass with no problem at all, all of which follow the exact same structure.
The only thing I will say to keep in mind is that in the folder "fogedretsmøde", the wave files are a different quality than those in the other folders. This was because I wanted to test, if that perhaps played a role, which it did not. Everything we have uploaded (and which has worked) was mono, 44khz 16 bit. I have tried to lower the khz on the files which fails, but it changes nothing.
Let me know if you need more info, and thanks for the assistance!
Best,
Jens
In addition, I have uploaded the json here with the failures. It refers to dataset: c26343d9-4492-46b1-bc3a-950bcd834dad
Hi,
I just wanted to ask if you have had any time to look at this? We are starting to get under a bit of pressure timewise, and I really need to find a solution for this.
Best,
Jens
![](data:image/svg+xml, %3Csvg xmlns='http://www.w3.org/2000/svg' height='64' class='font-weight-bold' style='font: 600 30.11764705882353px "SegoeUI", Arial' width='64'%3E%3Ccircle fill='hsl(89.60000000000001, 79%, 18%)' cx='32' cy='32' r='32' /%3E%3Ctext x='50%25' y='55%25' dominant-baseline='middle' text-anchor='middle' fill='%23FFF' %3EM%3C/text%3E%3C/svg%3E)
Had same issue and ticketed with support to solve. My problem was my blob was not set well. You can use Microsoft support to solve this issue.