Skip to content

Conversation

@Cadlaxa
Copy link
Contributor

@Cadlaxa Cadlaxa commented Feb 14, 2024

Custom English Arpabet Phonemizer based on Syllable-Based API Phonemizer with fallback support

  • Mechanics of this Phonemizer:

    • Starting C: [- c]
    • Starting CV: [- c v] or [- cv]
    • Starting CCV: [- ccv] or [- cc v]
    • Starting V: [- v] or [v]
    • VV: (Fallbacks to [v c][c v]/[cv] then [c v]/[cv] then [v], if vb doesn't have [v], glottal stop are used instead)
    • Connecting CV: [c v] (Fallbacks to [cv]if no alias is detected)
    • Connecting CCV: [ccv] or [cc v]
    • Connecting VC: [v c] then [v -] [- c
    • Connecting CC: [c c] (with consonant fallbacks) then [c1 -] [- c2]
    • Connecting CCC: [c cc] (fallbacks to [c c])
    • Ending C: [c -]
    • Ending V: [v -]
  • Phoneme length are specified directly to the phonemizer:

    • Default transition in ms: '1.0'
    • Vowels: 'default'
    • Consonants: '1.3'
    • Affricates: '1.5'
    • Long Consonants: '2.3'
    • Semi-long Consonants: '1.3'
    • Tap Consonant: '0.5'
    • Glide Consonants: '2.5'

Vowel and Consonant Fallbacks

  • This custom Phonemizer supports vowel and consonant fallbacks:
    • CV/CCV Fallback: [cc v] or [ccv] and [c v] or [cv]
    • VV Fallback: `([v c][c v]/[cv] then [c v]/[cv] then [v], if vb doesn't have [v], glottal stop are used instead)``
    • Connecting VC/VCC Fallback: [v c] then [v -] [- c]
    • Connecting CC/CCC: [c c] (with consonant fallbacks) then [c1 -] [- c2]
    • Starting and Ending Consonants

📍For more information, visit https://github.com/Cadlaxa/Syllable-Based-ARPAsing-Phonemizer

@stakira
Copy link
Owner

stakira commented Feb 25, 2024

The test you added seem to fail?

@stakira stakira merged commit e5456c3 into stakira:master Mar 28, 2024
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants