Azure AI Speech
An Azure service that integrates speech processing into apps and services.
1,555 questions
Non-Neural Legacy Training Availability
![](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(140.8, 14.000000000000002%, 23%)' cx='32' cy='32' r='32' /%3E%3Ctext x='50%25' y='55%25' dominant-baseline='middle' text-anchor='middle' fill='%23FFF' %3EAR%3C/text%3E%3C/svg%3E)
Andrew Read
1
Reputation point
I am attempting to train a new custom voice using one of the non-neural options. While I understand that both statistical parametric and concatenative are being completely discontinued in 2024, are there any regions were either of these options are still available? Given that custom neural is by request only and that it is significantly more expensive, it seems quite odd that the legacy options are unavailable during this transitional period. There appears to be a service hole in that custom neural isn't even widely available, nor price competitive.
Please advise.
Thanks!
{count} votes
-
romungi-MSFT 43,696 Reputation points Microsoft Employee2021-08-24T07:35:26.71+00:00 @Andrew Read As part of the upgrades to the services the recent announcement to migrate to neural voices have been announced. You can use the standard voices until the retirement in August 2024, the voices seem to be available from the speech studio too.
The availability of custom neural voices is actually GA since February 2021 as announced but customers need to still apply for usage. This is primarily for data privacy & security reasons to avoid misuse of the service.
With respect to the pricing and availability we will convey the feedback to the product group so they could review and take necessary actions. Thanks.
-
Andrew Read 1 Reputation point
2021-08-24T14:22:18.943+00:00 Thanks for the reply.
While I am aware and have used the existing voices, I am attempting to train a custom voice that uses my own voice as the template using the legacy options. Custom Neural is cost-prohibitive at the moment, mainly due to the $52/hr compute charge for training.
The only option I get presented with is:
While I am looking for the following:
-
romungi-MSFT 43,696 Reputation points Microsoft Employee
2021-08-26T09:18:40.647+00:00 @Andrew Read I believe you are seeing these options as unavailable while trying to train a model for custom voice. In which region is your speech resource mapped?
-
Andrew Read 1 Reputation point
2021-08-26T14:06:01.44+00:00 East US but I have tried several different regions with the same results.
Sign in to comment