storytold/storyteller-ml
2
0
Fork 0
mirror of https://github.com/storytold/storyteller-ml.git synced 2026-10-27 08:58:13 +00:00
Code Issues Packages Projects Releases Wiki Activity

Add sequencing, improvements to pipeline

Browse Source
This commit is contained in:
Brandon Thomas
2022-07-02 14:53:33 -04:00
parent 22a6986bd4
commit 57dfe576a8
13 changed files with 124 additions and 55 deletions
Download Patch File Download Diff File Expand all files Collapse all files
tts/text/phonetization/cmudict.py
+2
View File
@@ -4,6 +4,8 @@ Contents of this file are mostly from https://github.com/keithito/tacotron
import re
# TODO: Don't use strings for arpabet phoneme representation, and share the memory.
# Relative to TTS codebase root
# This does not include any CMUDict extensions or alternatives (eg. for other languages).
DEFAULT_CMUDICT_PATH = 'dictionaries/cmudict-0.7b'
tts/text/phonetization/english_phonetization.py
+6 -1
View File
@@ -7,7 +7,12 @@ class EnglishPhonetization(BasePhonetization):
def __init__(self, phoneme_dictionary: CMUDict):
super().__init__(phoneme_dictionary)
def convert_to_phonemes(self, grapheme_sequence: str) -> List[str]:
def try_convert_to_phonemes(self, grapheme_sequence: str) -> List[str]:
"""
Attempt to enrich a grapheme sequence with phonemes where word lookup succeeds.
Grapheme symbols will be left in place.
The output list will need to be space-separated before inference.
"""
words = self._split_grapheme_sentence_to_words(grapheme_sequence)
converted_words = []
tts/text/phonetization/test_english_phonetization.py
+6 -6
View File
@@ -6,27 +6,27 @@ class TestEnglishPhonetization:
def test_no_input(self):
phoneme_dictionary = CMUDict(DEFAULT_CMUDICT_PATH)
phonetization = EnglishPhonetization(phoneme_dictionary)
output = phonetization.convert_to_phonemes("")
output = phonetization.try_convert_to_phonemes("")
assert output == []
def test_all_words_in_dictionary(self):
phoneme_dictionary = CMUDict(DEFAULT_CMUDICT_PATH)
phonetization = EnglishPhonetization(phoneme_dictionary)
output = phonetization.convert_to_phonemes("this is a test")
output = phonetization.try_convert_to_phonemes("this is a test")
assert output == ["DH IH1 S", "IH1 Z", "AH0", "T EH1 S T"]
def test_most_words_in_dictionary(self):
phoneme_dictionary = CMUDict(DEFAULT_CMUDICT_PATH)
phonetization = EnglishPhonetization(phoneme_dictionary)
output = phonetization.convert_to_phonemes("nonword is a test")
output = phonetization.try_convert_to_phonemes("nonword is a test")
assert output == ["nonword", "IH1 Z", "AH0", "T EH1 S T"]
def test_punctuation(self):
phoneme_dictionary = CMUDict(DEFAULT_CMUDICT_PATH)
phonetization = EnglishPhonetization(phoneme_dictionary)
output = phonetization.convert_to_phonemes("this...")
output = phonetization.try_convert_to_phonemes("this...")
assert output == ["DH IH1 S", "..."]
output = phonetization.convert_to_phonemes("this!")
output = phonetization.try_convert_to_phonemes("this!")
assert output == ["DH IH1 S", "!"]
output = phonetization.convert_to_phonemes("this?")
output = phonetization.try_convert_to_phonemes("this?")
assert output == ["DH IH1 S", "?"]