My text-to-speech reader suddenly rises or drops in pitch whenever it hits certain words, breaking the flow for my son who listens to it every day. I need the voice to stay steady from start to finish, without the odd jumps that pop up mid-sentence. The current setup feeds plain text into a standard TTS engine, then exports to MP3. I’m open to whatever mix of model fine-tuning, configuration tweaks, or post-processing (Python, Audacity, Adobe Audition—your call) will smooth things out while keeping the voice natural. Deliverables • Updated settings, code, or processing chain that removes the pitch spikes • Short demo file using my sample text, showing consistent tone • Quick step-by-step so I can repeat the fix on future files Acceptance criteria: one full chapter read-through with no audible pitch jumps on any of the previously problematic words or phrases. Let me know how you plan to tackle it and how soon you can turn it around.