I’m continuing to get AI Studio Flash TTS 3.1 truncation issues, with this error:
[original: beyond::dependency::INVALID_ARGUMENT] TextToSpeechRecipeRunner received text output from model, but only audio output modality is supported.
It seems like the TTS model thinks it has to produce text instead of audio halfway through. This is happening literally in all generations of text. I am putting sample context and a speech block every time, I’m not putting in a scene but I don’t feel like I must. I am getting incredibly frustrated. I’m using Firefox on a Windows 11 PC. I have honestly had it up to here as I use this software to make character RP scenarios as the acting from the model can be very good, and most other TTS generators fail in this regard. But if this model will continue to degrade, and if 2.5 Pro and Flash are my only options but deliver inferior quality, I don’t know what to do! I’m furious!