Repository navigation
tts: say plainly that this is a dubbing default, not a narration one - #52
Merged
Merged
Conversation
mixing.text_to_speech is public and general-purpose, and its defaults are eleven_multilingual_v2 at stability 0.5. For DUBBING that is right: a dub tracks a performance that already exists, the timing is fixed by the source, and per-take variance is a defect rather than life. For NARRATION it is the wrong default, because v2 cannot render inline [audio tags] at all — and tag density is the main lever on whether a read sounds alive (plain text 70.8 Hz of pitch range, densely tagged 124.4 Hz). A narration built on this default is capped at the flat end before a word is written; one such render came back from a listener as 'somniferous'. So the defaults stay (flipping them would regress dubbing) and the docstrings now name the split and point narration work at braidio, which owns expressive delivery and defaults to eleven_v3. Docstrings only; no behaviour change. Suite 500 passed.
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
mixing.text_to_speechis public and general-purpose, and its defaults areeleven_multilingual_v2atstability0.5.For dubbing that is correct and stays unchanged: a dub tracks a performance that already exists, the timing is fixed by the source, the delivery belongs to the original actor, and per-take variance is a defect rather than life.
For narration it is the wrong default, because v2 cannot render inline
[audio tags]at all — and tag density is the single biggest lever on whether a read sounds alive (measured: plain text 70.8 Hz of pitch range, the same sentence densely tagged 124.4 Hz;stabilityacross its entire range moves it only 85 → 72 Hz). A narration built on this default is capped at the flat end before an author writes a word. One such render came back from a listener as somniferous.Flipping the default would regress dubbing, so it stays. What changes is that the docstrings now name the split and send narration/commentary work to
braidio, which owns expressive delivery and now defaults toeleven_v3(braidio#71).Docstrings only; no behaviour change. Suite 500 passed.