When using other providers for TTS, I can use the On Text Received function and the audio duration parameter to correctly show subtitles as the characters are speaking. However, with ElevenLabs the text just seems to come in immediately, and the duration is incorrect (Talking Time Remaining is also incorrect). This means it’s impossible to show subtitles correctly timed. They also seem to never be split up into sentences the way they are with other voices.
Would this be possible to fix? I’ve reported this issue on Discord before (Audio duration for subtitles in UE5), but the suggested fix doesn’t work, as that info is also incorrect.
Thank you for bringing this up, and we appreciate your patience.
ElevenLabs operates differently compared to other TTS providers, and unfortunately, this feature does not work as expected with ElevenLabs voices. At the moment, there isn’t a simple solution to accurately sync subtitles using the On Text Received function and audio duration with ElevenLabs generated speech.
In this case, implementing a custom solution may be necessary. You might explore available online resources or consider using AI-based approaches to approximate subtitle timing more accurately.
We understand that this is not an ideal response, but we value your feedback and will continue to look into ways to improve the integration. Let us know if you need further assistance!