Forum Discussion
Save settings in Text-to-Speech
Strong plus-one on the favorites idea from the production side. When you're voicing a multi-course series, even small drift in speed or stability settings between batches is audible — learners notice, reviewers definitely notice. Until favorites exist, the workaround that's held up for me is keeping a one-line 'voice profile' (voice name + the three settings) pinned to the top of the script doc, so every generation session starts from the same numbers. Not as good as saved presets, but it kills the drift.
Hello ronald-hicks,
Thanks for sharing this workaround! It's helpful to hear how even small differences in voice settings can affect the consistency of narration across your courses.
I'm curious:
- Would you see yourself reusing the same saved voice settings throughout a single course or across several courses?
- When narration sounds different between batches, how much rework does that typically create for you?
Appreciate you sharing your production experience and helping us better understand the impact!
Hi Luciana — across several courses, for sure. The use case I keep running into is a series: one voice profile across a whole curriculum, so lesson one and lesson eight sound like they were voiced in the same session. Within a single course it's handy too, but cross-course consistency is where saved presets would earn their keep.
On rework: when a batch drifts, it's regenerating the affected audio, then listening to every new take against the approved profile, plus re-checking slide timing wherever a new take runs longer or shorter. No single step is huge, but across a series it stacks up — and the painful version is a reviewer flagging that module six sounds like a different narrator.
Hi ronald-hicks,
Thank you for such a clear follow-up! I understand that saved presets would matter most across a series of courses, so lesson one and lesson eight sound like they were voiced in the same session. When a batch drifts, you regenerate the audio, compare each take against the approved profile, and re-check slide timing. That adds up, especially when a reviewer flags that a module sounds as if it were recorded by a different narrator.
We appreciate you sharing how this plays out in your production work. We can’t promise a specific outcome, but your details help us understand what you’re dealing with. If more comes up, we’d love to hear it!