mirror of
https://github.com/storytold/storyteller-ml.git
synced 2026-10-18 07:41:03 +00:00
Brandon Thomas
47 lines
1.5 KiB
Python
47 lines
1.5 KiB
Python
from enum import Enum, auto
|
|
|
|
from tts.text.input_parsing.parser import GraphemeSquence, parse_grapheme_and_arpabet_sequence
|
|
from tts.text.normalization.english_normalization import EnglishNormalization
|
|
from tts.text.normalization.spanish_normalization import SpanishNormalization
|
|
|
|
|
|
class PipelineType(Enum):
|
|
LEGACY_FAKEYOU = auto()
|
|
ENGLISH_FORCED_ARPABET_V1 = auto()
|
|
SPANISH_FORCED_ARPABET_V1 = auto()
|
|
|
|
_ENGLISH_NORMALIZER = EnglishNormalization()
|
|
_SPANISH_NORMALIZER = SpanishNormalization()
|
|
|
|
def _legacy_fakeyou_pipeline(input_text):
|
|
"""
|
|
Legacy FakeYou.com did not do Grapheme -> Arpabet conversions automatically.
|
|
"""
|
|
pass
|
|
|
|
|
|
def _english_forced_arpabet_v1(input_text):
|
|
parsed_sequence = parse_grapheme_and_arpabet_sequence(input_text)
|
|
for i in range(len(parsed_sequence)):
|
|
subsequence = parsed_sequence[i]
|
|
if isinstance(subsequence, GraphemeSquence):
|
|
normalized = _ENGLISH_NORMALIZER.normalize(subsequence.sequence)
|
|
parsed_sequence[i] = GraphemeSquence(normalized)
|
|
|
|
integer_sequence = []
|
|
|
|
|
|
|
|
def _spanish_forced_arpabet_v1(input_text):
|
|
parsed_sequence = parse_grapheme_and_arpabet_sequence(input_text)
|
|
|
|
|
|
def text_to_sequence(input_text: str, pipeline_type: PipelineType):
|
|
if pipeline_type == PipelineType.ENGLISH_FORCED_ARPABET_V1:
|
|
return _english_forced_arpabet_v1(input_text)
|
|
elif pipeline_type == PipelineType.SPANISH_FORCED_ARPABET_V1:
|
|
return _spanish_forced_arpabet_v1(input_text)
|
|
else:
|
|
return _legacy_fakeyou_pipeline(input_text)
|
|
|