TTS prompt
#57
by
anferico
- opened
What is a good prompt for text to speech (TTS)? I tried many different prompts, but there seems to be no way to prevent the model from generating additional audio at the end. Literally every time, the model ends the speech with things like "that's it. Would you like me to generate another speech?" or "what are your thoughts on this?" and similar things. I tried adding specific instructions to abstain from generating additional things (both in the system prompt and in the user prompt), but the model is ignoring them.
Here are some prompts I tried:
- Generate speech for the following text: ''
- Read this sentence out loud: ''
- Read this sentence out loud, then shut up: ''
- Speak the following sentence out loud: ''
- Generate speech based on the following text and please, do NOT generate anything else afterwards: ''
- I want you to generate speech from a given text. It is of utmost importance that you do not generate anything else afterwards, including but not limited to things like 'would you like to hear more about this topic?', 'would you like me to generate another speech?' and similar things. The text to generate speech from is: ''
This comment has been hidden