saysynth.lib.nlp

Utilities for extracing phonemes and syllables from english text.

View Source

  1"""
  2Utilities for extracing phonemes and syllables from english text.
  3"""
  4
  5from functools import lru_cache
  6from typing import List, Tuple
  7
  8import nltk
  9from g2p_en import G2p
 10from nltk.corpus import cmudict
 11from nltk.tokenize import word_tokenize
 12
 13# ensure NLTK resources are downloaded
 14try:
 15    cmudict.ensure_loaded()
 16except LookupError:
 17    nltk.download("cmudict")
 18try:
 19    word_tokenize("test")
 20except Exception:
 21    nltk.download("punkt")
 22
 23
 24from ..constants import G2P_PHONEMES_TO_SAY_PHONEMES
 25
 26G2P = None
 27CMU = None
 28VOWELS = "aeiouy"
 29
 30
 31@lru_cache(maxsize=1024)
 32def word_to_g2p_phonemes(text: str) -> List[str]:
 33    """
 34    Extract g2p phonemes from a text.
 35
 36    Args:
 37        text: A text to extract phonemes from
 38    """
 39    global G2P
 40    if not G2P:
 41        G2P = G2p()
 42    return G2P(text)
 43
 44
 45def word_to_say_phonemes(text: str) -> List[str]:
 46    """
 47    Extract Apple SpeechSynthesis phonemes from a text.
 48
 49    Args:
 50        text: A text to extract phonemes from
 51    """
 52    return [
 53        G2P_PHONEMES_TO_SAY_PHONEMES.get(p, "")
 54        for p in word_to_g2p_phonemes(text)
 55    ]
 56
 57
 58@lru_cache(maxsize=1024)
 59def word_to_syllable_count(word: str) -> int:
 60    """
 61    Compute the syllable count of a word, using CMU if the word is present in the
 62    corpus, otherwise [follow this implementation](https://stackoverflow.com/questions/14541303/count-the-number-of-syllables-in-a-word).
 63
 64    Args:
 65        word: A word to compute syllables for.
 66    """
 67    global CMU
 68    if not CMU:
 69        CMU = cmudict.dict()
 70    try:
 71        return [
 72            len(list(y for y in x if y[-1].isdigit()))
 73            for x in CMU[word.lower()]
 74        ][0]
 75    except KeyError:
 76        # if word not found in cmudict
 77        # referred from stackoverflow.com/questions/14541303/count-the-number-of-syllables-in-a-word
 78        count = 0
 79        word = word.lower()
 80        if word[0] in VOWELS:
 81            count += 1
 82        for index in range(1, len(word)):
 83            if word[index] in VOWELS and word[index - 1] not in VOWELS:
 84                count += 1
 85        if word.endswith("e"):
 86            count -= 1
 87        if word.endswith("le"):
 88            count += 1
 89        if count == 0:
 90            count += 1
 91        return count
 92
 93
 94@lru_cache(maxsize=1024)
 95def process_text_for_say(text: str) -> List[Tuple[str, int, List[str]]]:
 96    """
 97    Get a list of phonemes + syllable counts for each word in a text.
 98    Args:
 99        text: A text to process.
100    """
101    return [
102        (word, word_to_syllable_count(word), word_to_say_phonemes(word))
103        if word not in [",", ".", "?", "!", "-", ":", ";"]
104        else ("", 1, ["%"])  # silence
105        for word in word_tokenize(text)
106    ]

@lru_cache(maxsize=1024)

def word_to_g2p_phonemes(text: str) -> List[str]: View Source

32@lru_cache(maxsize=1024)
33def word_to_g2p_phonemes(text: str) -> List[str]:
34    """
35    Extract g2p phonemes from a text.
36
37    Args:
38        text: A text to extract phonemes from
39    """
40    global G2P
41    if not G2P:
42        G2P = G2p()
43    return G2P(text)

Extract g2p phonemes from a text.

Arguments:

text: A text to extract phonemes from

def word_to_say_phonemes(text: str) -> List[str]: View Source

46def word_to_say_phonemes(text: str) -> List[str]:
47    """
48    Extract Apple SpeechSynthesis phonemes from a text.
49
50    Args:
51        text: A text to extract phonemes from
52    """
53    return [
54        G2P_PHONEMES_TO_SAY_PHONEMES.get(p, "")
55        for p in word_to_g2p_phonemes(text)
56    ]

Extract Apple SpeechSynthesis phonemes from a text.

Arguments:

text: A text to extract phonemes from

@lru_cache(maxsize=1024)

def word_to_syllable_count(word: str) -> int: View Source

59@lru_cache(maxsize=1024)
60def word_to_syllable_count(word: str) -> int:
61    """
62    Compute the syllable count of a word, using CMU if the word is present in the
63    corpus, otherwise [follow this implementation](https://stackoverflow.com/questions/14541303/count-the-number-of-syllables-in-a-word).
64
65    Args:
66        word: A word to compute syllables for.
67    """
68    global CMU
69    if not CMU:
70        CMU = cmudict.dict()
71    try:
72        return [
73            len(list(y for y in x if y[-1].isdigit()))
74            for x in CMU[word.lower()]
75        ][0]
76    except KeyError:
77        # if word not found in cmudict
78        # referred from stackoverflow.com/questions/14541303/count-the-number-of-syllables-in-a-word
79        count = 0
80        word = word.lower()
81        if word[0] in VOWELS:
82            count += 1
83        for index in range(1, len(word)):
84            if word[index] in VOWELS and word[index - 1] not in VOWELS:
85                count += 1
86        if word.endswith("e"):
87            count -= 1
88        if word.endswith("le"):
89            count += 1
90        if count == 0:
91            count += 1
92        return count

Compute the syllable count of a word, using CMU if the word is present in the corpus, otherwise follow this implementation.

Arguments:

word: A word to compute syllables for.

@lru_cache(maxsize=1024)

def process_text_for_say(text: str) -> List[Tuple[str, int, List[str]]]: View Source

 95@lru_cache(maxsize=1024)
 96def process_text_for_say(text: str) -> List[Tuple[str, int, List[str]]]:
 97    """
 98    Get a list of phonemes + syllable counts for each word in a text.
 99    Args:
100        text: A text to process.
101    """
102    return [
103        (word, word_to_syllable_count(word), word_to_say_phonemes(word))
104        if word not in [",", ".", "?", "!", "-", ":", ";"]
105        else ("", 1, ["%"])  # silence
106        for word in word_tokenize(text)
107    ]

Get a list of phonemes + syllable counts for each word in a text.

Arguments:

text: A text to process.