Thank you for your thorough testing of the S2ST Beta and for the continuous feedback youāve provided. Below is a consolidated response to the issues and suggestions you raised. We also encourage continued evaluations focusing on the actual output quality and product behavior.
About the 5-minute duration limit (Beta explanation)
The current 5-minute limit in the Beta is designed to reduce server resource usage so that more users can experience the feature simultaneously. It also helps avoid long processing times when testing full-length videos, which would slow down validation of the output quality.
The official release will not include any duration limit.
In an upcoming Beta update, we plan to introduce automatic processing of the first 5 minutes of any video, so no manual trimming will be required.
To achieve the best voice cloning performance, we recommend using source videos with accurate subtitles. Subtitles can be either embedded tracks or external .srt files. Also, please ensure that the selected audio language and subtitle language in the client are correct before processing.
About subtitle output logic
The current version supports both Remux output and exporting subtitles as standalone SRT files.
In the next update, we will add a feature to remember the userās last-selected output format. For example, if you choose SRT, the system will default to SRT for future tasks. This will make repeated or batch processing more efficient.
About customizable audio codec formats (DTS / EAC3)
In the next version, we will introduce an Audio Upmix feature and support exporting audio using formats such as DTS and EAC3. This aims to improve the experience for advanced users and home theater environments.
About the billing model
Pure local processing is not supported during the Beta phase.
When the official version launches, you will be able to choose a subscription plan based on your usage needs.
About lip-sync
Lip-sync capability has been included in our long-term planning.
For now, our focus remains on improving translation accuracy, voice cloning naturalness, voice consistency across characters, and multi-language processing performance. We welcome continued feedback in these areas.
Thank you again for your valuable input. Your testing is extremely important to our product improvement. Please feel free to reach out with any additional suggestions or questions.
