Now that Piper TTS has made it to iOS and is relatively stable on newer devices, I'd like to create a Community hub of sorts, where people can upload voices they've created for it.
-
Now that Piper TTS has made it to iOS and is relatively stable on newer devices, I'd like to create a Community hub of sorts, where people can upload voices they've created for it. To go along with this, I'd like to release some kind of training utility to make the process a little less scary for people who don't code regularly. Is this something you would be interested in?
My goal with this is to release more high-quality voices than what's currently available, many of the issues people see with Piper simply come down to the data used for training not being up to a professional standard. Honestly I think the main issue will be data collection, I have absolutely no interest in scraping from the web or using unlicensed data to train like many of these other open source TTS projects.
-
Now that Piper TTS has made it to iOS and is relatively stable on newer devices, I'd like to create a Community hub of sorts, where people can upload voices they've created for it. To go along with this, I'd like to release some kind of training utility to make the process a little less scary for people who don't code regularly. Is this something you would be interested in?
My goal with this is to release more high-quality voices than what's currently available, many of the issues people see with Piper simply come down to the data used for training not being up to a professional standard. Honestly I think the main issue will be data collection, I have absolutely no interest in scraping from the web or using unlicensed data to train like many of these other open source TTS projects.
@ZBennoui Wait, has it been fixed? I could not import my own local voice files last time I tried, and it had considerable issues.
-
@ZBennoui Wait, has it been fixed? I could not import my own local voice files last time I tried, and it had considerable issues.
@BorrisInABox So from talking with @bscross32 it seems like the issue relates to when the models were trained. Piper underwent some changes after the version for NVDA was abandoned, and I believe the iOS version is using the latest release or close to it. I'll need to try training up a new voice over the next few days and test to make sure it works.
-
@BorrisInABox So from talking with @bscross32 it seems like the issue relates to when the models were trained. Piper underwent some changes after the version for NVDA was abandoned, and I believe the iOS version is using the latest release or close to it. I'll need to try training up a new voice over the next few days and test to make sure it works.
@ZBennoui @BorrisInABox I think so, it was down to the voice models not having every single field in the Json that the iOS app expected.
-
@BorrisInABox So from talking with @bscross32 it seems like the issue relates to when the models were trained. Piper underwent some changes after the version for NVDA was abandoned, and I believe the iOS version is using the latest release or close to it. I'll need to try training up a new voice over the next few days and test to make sure it works.
@ZBennoui @bscross32 I don't know much other than I could load these voices on the latest Piper on Linux, but not on iOS.
-
R relay@relay.mycrowd.ca shared this topicR relay@relay.infosec.exchange shared this topicR relay@relay.publicsquare.global shared this topic
