We are aware of an issue where users were unable to generate voice transcripts starting on Friday, October 2. This was due to a failed configuration change that was applied to one of our cloud resources used for this feature. This was immediately identified and was thought to have been fully rolled back, but in reality, the configuration was only partially rolled back. We have since fully rolled back the configuration change and have verified that voice transcript generation is working again.
We are in the process of identifying the sequence of events that led to this so that we can apply mitigation strategies to prevent similar incidents in the future. We apologize for any inconvenience this may have caused.