I have trained a text-to-speech engine to synthesize plain tone speech pretty naturally. So now I would like to expand its ability and let it ask people questions, or stress certain words in its answer. However re-train the whole TTS engine with different tones is very hard to implement, as you will need to control the voice actors' tones very carefully, and the current neural engine doesn't do this kind of magic.

So my question is, is it possible that I can use the TTS to synthesize the audio in plain tone, then I use modern DAWs or other tricks to make it into a more suitable speech?

For example from TTS I can get "How are you today", with all 4 words being totally flat and mechanic. Then I may stress the word "How" and lift "you", making it a proper greeting that significantly improves user experience.

Is it possible? If so how could I get started?

New contributor
JimmyK is a new contributor to this site. Take care in asking for clarification, commenting, and answering. Check out our Code of Conduct.
  • It's trivial doing this sort of thing in a DAW, however I fear that adding a DAW into the workflow will degrade the overall user experience to a greater degree that by leaving things as they are! I fear your only way to do this efficiently, and as part of the TTS process, is to modify the neural engine and to include some pitch manipulation into the overall synthesis layer. – Mark Dec 5 at 14:19

Your Answer

JimmyK is a new contributor. Be nice, and check out our Code of Conduct.

By clicking "Post Your Answer", you acknowledge that you have read our updated terms of service, privacy policy and cookie policy, and that your continued use of the website is subject to these policies.

Browse other questions tagged or ask your own question.