rename

melo/text/fr_phonemizer/base.py

@@ -0,0 +1,140 @@
+import abc
+from typing import List, Tuple
+
+from .punctuation import Punctuation
+
+
+class BasePhonemizer(abc.ABC):
+    """Base phonemizer class
+
+    Phonemization follows the following steps:
+        1. Preprocessing:
+            - remove empty lines
+            - remove punctuation
+            - keep track of punctuation marks
+
+        2. Phonemization:
+            - convert text to phonemes
+
+        3. Postprocessing:
+            - join phonemes
+            - restore punctuation marks
+
+    Args:
+        language (str):
+            Language used by the phonemizer.
+
+        punctuations (List[str]):
+            List of punctuation marks to be preserved.
+
+        keep_puncs (bool):
+            Whether to preserve punctuation marks or not.
+    """
+
+    def __init__(self, language, punctuations=Punctuation.default_puncs(), keep_puncs=False):
+        # ensure the backend is installed on the system
+        if not self.is_available():
+            raise RuntimeError("{} not installed on your system".format(self.name()))  # pragma: nocover
+
+        # ensure the backend support the requested language
+        self._language = self._init_language(language)
+
+        # setup punctuation processing
+        self._keep_puncs = keep_puncs
+        self._punctuator = Punctuation(punctuations)
+
+    def _init_language(self, language):
+        """Language initialization
+
+        This method may be overloaded in child classes (see Segments backend)
+
+        """
+        if not self.is_supported_language(language):
+            raise RuntimeError(f'language "{language}" is not supported by the ' f"{self.name()} backend")
+        return language
+
+    @property
+    def language(self):
+        """The language code configured to be used for phonemization"""
+        return self._language
+
+    @staticmethod
+    @abc.abstractmethod
+    def name():
+        """The name of the backend"""
+        ...
+
+    @classmethod
+    @abc.abstractmethod
+    def is_available(cls):
+        """Returns True if the backend is installed, False otherwise"""
+        ...
+
+    @classmethod
+    @abc.abstractmethod
+    def version(cls):
+        """Return the backend version as a tuple (major, minor, patch)"""
+        ...
+
+    @staticmethod
+    @abc.abstractmethod
+    def supported_languages():
+        """Return a dict of language codes -> name supported by the backend"""
+        ...
+
+    def is_supported_language(self, language):
+        """Returns True if `language` is supported by the backend"""
+        return language in self.supported_languages()
+
+    @abc.abstractmethod
+    def _phonemize(self, text, separator):
+        """The main phonemization method"""
+
+    def _phonemize_preprocess(self, text) -> Tuple[List[str], List]:
+        """Preprocess the text before phonemization
+
+        1. remove spaces
+        2. remove punctuation
+
+        Override this if you need a different behaviour
+        """
+        text = text.strip()
+        if self._keep_puncs:
+            # a tuple (text, punctuation marks)
+            return self._punctuator.strip_to_restore(text)
+        return [self._punctuator.strip(text)], []
+
+    def _phonemize_postprocess(self, phonemized, punctuations) -> str:
+        """Postprocess the raw phonemized output
+
+        Override this if you need a different behaviour
+        """
+        if self._keep_puncs:
+            return self._punctuator.restore(phonemized, punctuations)[0]
+        return phonemized[0]
+
+    def phonemize(self, text: str, separator="|", language: str = None) -> str:  # pylint: disable=unused-argument
+        """Returns the `text` phonemized for the given language
+
+        Args:
+            text (str):
+                Text to be phonemized.
+
+            separator (str):
+                string separator used between phonemes. Default to '_'.
+
+        Returns:
+            (str): Phonemized text
+        """
+        text, punctuations = self._phonemize_preprocess(text)
+        phonemized = []
+        for t in text:
+            p = self._phonemize(t, separator)
+            phonemized.append(p)
+        phonemized = self._phonemize_postprocess(phonemized, punctuations)
+        return phonemized
+
+    def print_logs(self, level: int = 0):
+        indent = "\t" * level
+        print(f"{indent}| > phoneme language: {self.language}")
+        print(f"{indent}| > phoneme backend: {self.name()}")

melo/text/fr_phonemizer/cleaner.py

@@ -0,0 +1,122 @@
+"""Set of default text cleaners"""
+# TODO: pick the cleaner for languages dynamically
+
+import re
+from .french_abbreviations import abbreviations_fr
+
+# Regular expression matching whitespace:
+_whitespace_re = re.compile(r"\s+")
+
+
+rep_map = {
+    "：": ",",
+    "；": ",",
+    "，": ",",
+    "。": ".",
+    "！": "!",
+    "？": "?",
+    "\n": ".",
+    "·": ",",
+    "、": ",",
+    "...": ".",
+    "…": ".",
+    "$": ".",
+    "“": "",
+    "”": "",
+    "‘": "",
+    "’": "",
+    "（": "",
+    "）": "",
+    "(": "",
+    ")": "",
+    "《": "",
+    "》": "",
+    "【": "",
+    "】": "",
+    "[": "",
+    "]": "",
+    "—": "",
+    "～": "-",
+    "~": "-",
+    "「": "",
+    "」": "",
+    "¿" : "",
+    "¡" : ""
+}
+
+
+def replace_punctuation(text):
+    pattern = re.compile("|".join(re.escape(p) for p in rep_map.keys()))
+    replaced_text = pattern.sub(lambda x: rep_map[x.group()], text)
+    return replaced_text
+
+def expand_abbreviations(text, lang="fr"):
+    if lang == "fr":
+        _abbreviations = abbreviations_fr
+    for regex, replacement in _abbreviations:
+        text = re.sub(regex, replacement, text)
+    return text
+
+
+def lowercase(text):
+    return text.lower()
+
+
+def collapse_whitespace(text):
+    return re.sub(_whitespace_re, " ", text).strip()
+
+def remove_punctuation_at_begin(text):
+    return re.sub(r'^[,.!?]+', '', text)
+
+def remove_aux_symbols(text):
+    text = re.sub(r"[\<\>\(\)\[\]\"\«\»]+", "", text)
+    return text
+
+
+def replace_symbols(text, lang="en"):
+    """Replace symbols based on the lenguage tag.
+
+    Args:
+      text:
+       Input text.
+      lang:
+        Lenguage identifier. ex: "en", "fr", "pt", "ca".
+
+    Returns:
+      The modified text
+      example:
+        input args:
+            text: "si l'avi cau, diguem-ho"
+            lang: "ca"
+        Output:
+            text: "si lavi cau, diguemho"
+    """
+    text = text.replace(";", ",")
+    text = text.replace("-", " ") if lang != "ca" else text.replace("-", "")
+    text = text.replace(":", ",")
+    if lang == "en":
+        text = text.replace("&", " and ")
+    elif lang == "fr":
+        text = text.replace("&", " et ")
+    elif lang == "pt":
+        text = text.replace("&", " e ")
+    elif lang == "ca":
+        text = text.replace("&", " i ")
+        text = text.replace("'", "")
+    elif lang== "es":
+        text=text.replace("&","y")
+        text = text.replace("'", "")
+    return text
+
+def french_cleaners(text):
+    """Pipeline for French text. There is no need to expand numbers, phonemizer already does that"""
+    text = expand_abbreviations(text, lang="fr")
+    # text = lowercase(text) # as we use the cased bert
+    text = replace_punctuation(text)
+    text = replace_symbols(text, lang="fr")
+    text = remove_aux_symbols(text)
+    text = remove_punctuation_at_begin(text)
+    text = collapse_whitespace(text)
+    text = re.sub(r'([^\.,!\?\-…])$', r'\1.', text)
+    return text
+

melo/text/fr_phonemizer/en_symbols.json

@@ -0,0 +1,78 @@
+{"symbols": [
+    "_",
+    ",",
+    ".",
+    "!",
+    "?",
+    "-",
+    "~",
+    "\u2026",
+    "N",
+    "Q",
+    "a",
+    "b",
+    "d",
+    "e",
+    "f",
+    "g",
+    "h",
+    "i",
+    "j",
+    "k",
+    "l",
+    "m",
+    "n",
+    "o",
+    "p",
+    "s",
+    "t",
+    "u",
+    "v",
+    "w",
+    "x",
+    "y",
+    "z",
+    "\u0251",
+    "\u00e6",
+    "\u0283",
+    "\u0291",
+    "\u00e7",
+    "\u026f",
+    "\u026a",
+    "\u0254",
+    "\u025b",
+    "\u0279",
+    "\u00f0",
+    "\u0259",
+    "\u026b",
+    "\u0265",
+    "\u0278",
+    "\u028a",
+    "\u027e",
+    "\u0292",
+    "\u03b8",
+    "\u03b2",
+    "\u014b",
+    "\u0266",
+    "\u207c",
+    "\u02b0",
+    "`",
+    "^",
+    "#",
+    "*",
+    "=",
+    "\u02c8",
+    "\u02cc",
+    "\u2192",
+    "\u2193",
+    "\u2191",
+    " ",
+    "ɣ",
+    "ɡ", 
+    "r", 
+    "ɲ", 
+    "ʝ", 
+    "ʎ",
+    "ː"
+  ]
+}

melo/text/fr_phonemizer/fr_symbols.json

@@ -0,0 +1,89 @@
+{
+    "symbols": [
+        "_",
+        ",",
+        ".",
+        "!",
+        "?",
+        "-",
+        "~",
+        "\u2026",
+        "N",
+        "Q",
+        "a",
+        "b",
+        "d",
+        "e",
+        "f",
+        "g",
+        "h",
+        "i",
+        "j",
+        "k",
+        "l",
+        "m",
+        "n",
+        "o",
+        "p",
+        "s",
+        "t",
+        "u",
+        "v",
+        "w",
+        "x",
+        "y",
+        "z",
+        "\u0251",
+        "\u00e6",
+        "\u0283",
+        "\u0291",
+        "\u00e7",
+        "\u026f",
+        "\u026a",
+        "\u0254",
+        "\u025b",
+        "\u0279",
+        "\u00f0",
+        "\u0259",
+        "\u026b",
+        "\u0265",
+        "\u0278",
+        "\u028a",
+        "\u027e",
+        "\u0292",
+        "\u03b8",
+        "\u03b2",
+        "\u014b",
+        "\u0266",
+        "\u207c",
+        "\u02b0",
+        "`",
+        "^",
+        "#",
+        "*",
+        "=",
+        "\u02c8",
+        "\u02cc",
+        "\u2192",
+        "\u2193",
+        "\u2191",
+        " ",
+        "\u0263",
+        "\u0261",
+        "r",
+        "\u0272",
+        "\u029d",
+        "\u028e",
+        "\u02d0",
+        
+        "\u0303",
+        "\u0153",
+        "\u00f8",
+        "\u0281",
+        "\u0252",
+        "\u028c",
+        "\u2014",
+        "\u025c",
+        "\u0250"
+    ]
+}

melo/text/fr_phonemizer/fr_to_ipa.py

@@ -0,0 +1,30 @@
+from .cleaner import french_cleaners
+from .gruut_wrapper import Gruut
+
+
+def remove_consecutive_t(input_str):
+    result = []
+    count = 0
+
+    for char in input_str:
+        if char == 't':
+            count += 1
+        else:
+            if count < 3:  
+                result.extend(['t'] * count)
+            count = 0
+            result.append(char)
+
+    if count < 3:
+        result.extend(['t'] * count)
+
+    return ''.join(result)
+
+def fr2ipa(text):
+    e = Gruut(language="fr-fr", keep_puncs=True, keep_stress=True, use_espeak_phonemes=True)
+    # text = french_cleaners(text)
+    phonemes = e.phonemize(text, separator="")
+    # print(phonemes)
+    phonemes = remove_consecutive_t(phonemes)
+    # print(phonemes)
+    return phonemes

melo/text/fr_phonemizer/french_abbreviations.py

@@ -0,0 +1,48 @@
+import re
+
+# List of (regular expression, replacement) pairs for abbreviations in french:
+abbreviations_fr = [
+    (re.compile("\\b%s\\." % x[0], re.IGNORECASE), x[1])
+    for x in [
+        ("M", "monsieur"),
+        ("Mlle", "mademoiselle"),
+        ("Mlles", "mesdemoiselles"),
+        ("Mme", "Madame"),
+        ("Mmes", "Mesdames"),
+        ("N.B", "nota bene"),
+        ("M", "monsieur"),
+        ("p.c.q", "parce que"),
+        ("Pr", "professeur"),
+        ("qqch", "quelque chose"),
+        ("rdv", "rendez-vous"),
+        ("max", "maximum"),
+        ("min", "minimum"),
+        ("no", "numéro"),
+        ("adr", "adresse"),
+        ("dr", "docteur"),
+        ("st", "saint"),
+        ("co", "companie"),
+        ("jr", "junior"),
+        ("sgt", "sergent"),
+        ("capt", "capitain"),
+        ("col", "colonel"),
+        ("av", "avenue"),
+        ("av. J.-C", "avant Jésus-Christ"),
+        ("apr. J.-C", "après Jésus-Christ"),
+        ("art", "article"),
+        ("boul", "boulevard"),
+        ("c.-à-d", "c’est-à-dire"),
+        ("etc", "et cetera"),
+        ("ex", "exemple"),
+        ("excl", "exclusivement"),
+        ("boul", "boulevard"),
+    ]
+] + [
+    (re.compile("\\b%s" % x[0]), x[1])
+    for x in [
+        ("Mlle", "mademoiselle"),
+        ("Mlles", "mesdemoiselles"),
+        ("Mme", "Madame"),
+        ("Mmes", "Mesdames"),
+    ]
+]

melo/text/fr_phonemizer/french_symbols.txt

@@ -0,0 +1 @@
+_,.!?-~…NQabdefghijklmnopstuvwxyzɑæʃʑçɯɪɔɛɹðəɫɥɸʊɾʒθβŋɦ⁼ʰ`^#*=ˈˌ→↓↑ ɣɡrɲʝʎː̃œøʁɒʌ—ɜɐ

melo/text/fr_phonemizer/gruut_wrapper.py

@@ -0,0 +1,258 @@
+import importlib
+from typing import List
+
+import gruut
+from gruut_ipa import IPA # pip install gruut_ipa
+
+from .base import BasePhonemizer
+from .punctuation import Punctuation
+
+# Table for str.translate to fix gruut/TTS phoneme mismatch
+GRUUT_TRANS_TABLE = str.maketrans("g", "ɡ")
+
+
+class Gruut(BasePhonemizer):
+    """Gruut wrapper for G2P
+
+    Args:
+        language (str):
+            Valid language code for the used backend.
+
+        punctuations (str):
+            Characters to be treated as punctuation. Defaults to `Punctuation.default_puncs()`.
+
+        keep_puncs (bool):
+            If true, keep the punctuations after phonemization. Defaults to True.
+
+        use_espeak_phonemes (bool):
+            If true, use espeak lexicons instead of default Gruut lexicons. Defaults to False.
+
+        keep_stress (bool):
+            If true, keep the stress characters after phonemization. Defaults to False.
+
+    Example:
+
+        >>> from TTS.tts.utils.text.phonemizers.gruut_wrapper import Gruut
+        >>> phonemizer = Gruut('en-us')
+        >>> phonemizer.phonemize("Be a voice, not an! echo?", separator="|")
+        'b|i| ə| v|ɔ|ɪ|s, n|ɑ|t| ə|n! ɛ|k|o|ʊ?'
+    """
+
+    def __init__(
+        self,
+        language: str,
+        punctuations=Punctuation.default_puncs(),
+        keep_puncs=True,
+        use_espeak_phonemes=False,
+        keep_stress=False,
+    ):
+        super().__init__(language, punctuations=punctuations, keep_puncs=keep_puncs)
+        self.use_espeak_phonemes = use_espeak_phonemes
+        self.keep_stress = keep_stress
+
+    @staticmethod
+    def name():
+        return "gruut"
+
+    def phonemize_gruut(self, text: str, separator: str = "|", tie=False) -> str:  # pylint: disable=unused-argument
+        """Convert input text to phonemes.
+
+        Gruut phonemizes the given `str` by seperating each phoneme character with `separator`, even for characters
+        that constitude a single sound.
+
+        It doesn't affect 🐸TTS since it individually converts each character to token IDs.
+
+        Examples::
+            "hello how are you today?" -> `h|ɛ|l|o|ʊ| h|a|ʊ| ɑ|ɹ| j|u| t|ə|d|e|ɪ`
+
+        Args:
+            text (str):
+                Text to be converted to phonemes.
+
+            tie (bool, optional) : When True use a '͡' character between
+                consecutive characters of a single phoneme. Else separate phoneme
+                with '_'. This option requires espeak>=1.49. Default to False.
+        """
+        ph_list = []
+        for sentence in gruut.sentences(text, lang=self.language, espeak=self.use_espeak_phonemes):
+            for word in sentence:
+                if word.is_break:
+                    # Use actual character for break phoneme (e.g., comma)
+                    if ph_list:
+                        # Join with previous word
+                        ph_list[-1].append(word.text)
+                    else:
+                        # First word is punctuation
+                        ph_list.append([word.text])
+                elif word.phonemes:
+                    # Add phonemes for word
+                    word_phonemes = []
+
+                    for word_phoneme in word.phonemes:
+                        if not self.keep_stress:
+                            # Remove primary/secondary stress
+                            word_phoneme = IPA.without_stress(word_phoneme)
+
+                        word_phoneme = word_phoneme.translate(GRUUT_TRANS_TABLE)
+
+                        if word_phoneme:
+                            # Flatten phonemes
+                            word_phonemes.extend(word_phoneme)
+
+                    if word_phonemes:
+                        ph_list.append(word_phonemes)
+
+        ph_words = [separator.join(word_phonemes) for word_phonemes in ph_list]
+        ph = f"{separator} ".join(ph_words)
+        return ph
+
+    def _phonemize(self, text, separator):
+        return self.phonemize_gruut(text, separator, tie=False)
+
+    def is_supported_language(self, language):
+        """Returns True if `language` is supported by the backend"""
+        return gruut.is_language_supported(language)
+
+    @staticmethod
+    def supported_languages() -> List:
+        """Get a dictionary of supported languages.
+
+        Returns:
+            List: List of language codes.
+        """
+        return list(gruut.get_supported_languages())
+
+    def version(self):
+        """Get the version of the used backend.
+
+        Returns:
+            str: Version of the used backend.
+        """
+        return gruut.__version__
+
+    @classmethod
+    def is_available(cls):
+        """Return true if ESpeak is available else false"""
+        return importlib.util.find_spec("gruut") is not None
+
+
+if __name__ == "__main__":
+    from cleaner import french_cleaners
+    import json
+
+    e = Gruut(language="fr-fr", keep_puncs=True, keep_stress=True, use_espeak_phonemes=True)
+    symbols = [  # en + sp
+        "_",
+        ",",
+        ".",
+        "!",
+        "?",
+        "-",
+        "~",
+        "\u2026",
+        "N",
+        "Q",
+        "a",
+        "b",
+        "d",
+        "e",
+        "f",
+        "g",
+        "h",
+        "i",
+        "j",
+        "k",
+        "l",
+        "m",
+        "n",
+        "o",
+        "p",
+        "s",
+        "t",
+        "u",
+        "v",
+        "w",
+        "x",
+        "y",
+        "z",
+        "\u0251",
+        "\u00e6",
+        "\u0283",
+        "\u0291",
+        "\u00e7",
+        "\u026f",
+        "\u026a",
+        "\u0254",
+        "\u025b",
+        "\u0279",
+        "\u00f0",
+        "\u0259",
+        "\u026b",
+        "\u0265",
+        "\u0278",
+        "\u028a",
+        "\u027e",
+        "\u0292",
+        "\u03b8",
+        "\u03b2",
+        "\u014b",
+        "\u0266",
+        "\u207c",
+        "\u02b0",
+        "`",
+        "^",
+        "#",
+        "*",
+        "=",
+        "\u02c8",
+        "\u02cc",
+        "\u2192",
+        "\u2193",
+        "\u2191",
+        " ",
+        "ɣ",
+        "ɡ", 
+        "r", 
+        "ɲ", 
+        "ʝ", 
+        "ʎ",
+        "ː"
+    ]
+    with open('/home/xumin/workspace/VITS-Training-Multiling/230715_fr/metadata.txt', 'r') as f:
+        lines = f.readlines()
+    
+
+    used_sym = []
+    not_existed_sym = []
+    phonemes = []
+
+    for line in lines:
+        text = line.split('|')[-1].strip()
+        text = french_cleaners(text)
+        ipa =  e.phonemize(text, separator="")
+        phonemes.append(ipa)
+        for s in ipa:
+            if s not in symbols:
+                if s not in not_existed_sym:
+                    print(f'not_existed char: {s}')
+                    not_existed_sym.append(s)
+            else:
+                if s not in used_sym:
+                    # print(f'used char: {s}')
+                    used_sym.append(s)
+    
+    print(used_sym)
+    print(not_existed_sym)
+
+
+    with open('./text/fr_phonemizer/french_symbols.txt', 'w') as g:
+        g.writelines(symbols + not_existed_sym)
+        
+    with open('./text/fr_phonemizer/example_ipa.txt', 'w') as g:
+        g.writelines(phonemes)
+
+    data = {'symbols': symbols + not_existed_sym}
+
+    with open('./text/fr_phonemizer/fr_symbols.json', 'w') as f:
+        json.dump(data, f, indent=4)
+

melo/text/fr_phonemizer/punctuation.py

@@ -0,0 +1,172 @@
+import collections
+import re
+from enum import Enum
+
+import six
+
+_DEF_PUNCS = ';:,.!?¡¿—…"«»“”'
+
+_PUNC_IDX = collections.namedtuple("_punc_index", ["punc", "position"])
+
+
+class PuncPosition(Enum):
+    """Enum for the punctuations positions"""
+
+    BEGIN = 0
+    END = 1
+    MIDDLE = 2
+    ALONE = 3
+
+
+class Punctuation:
+    """Handle punctuations in text.
+
+    Just strip punctuations from text or strip and restore them later.
+
+    Args:
+        puncs (str): The punctuations to be processed. Defaults to `_DEF_PUNCS`.
+
+    Example:
+        >>> punc = Punctuation()
+        >>> punc.strip("This is. example !")
+        'This is example'
+
+        >>> text_striped, punc_map = punc.strip_to_restore("This is. example !")
+        >>> ' '.join(text_striped)
+        'This is example'
+
+        >>> text_restored = punc.restore(text_striped, punc_map)
+        >>> text_restored[0]
+        'This is. example !'
+    """
+
+    def __init__(self, puncs: str = _DEF_PUNCS):
+        self.puncs = puncs
+
+    @staticmethod
+    def default_puncs():
+        """Return default set of punctuations."""
+        return _DEF_PUNCS
+
+    @property
+    def puncs(self):
+        return self._puncs
+
+    @puncs.setter
+    def puncs(self, value):
+        if not isinstance(value, six.string_types):
+            raise ValueError("[!] Punctuations must be of type str.")
+        self._puncs = "".join(list(dict.fromkeys(list(value))))  # remove duplicates without changing the oreder
+        self.puncs_regular_exp = re.compile(rf"(\s*[{re.escape(self._puncs)}]+\s*)+")
+
+    def strip(self, text):
+        """Remove all the punctuations by replacing with `space`.
+
+        Args:
+            text (str): The text to be processed.
+
+        Example::
+
+            "This is. example !" -> "This is example "
+        """
+        return re.sub(self.puncs_regular_exp, " ", text).rstrip().lstrip()
+
+    def strip_to_restore(self, text):
+        """Remove punctuations from text to restore them later.
+
+        Args:
+            text (str): The text to be processed.
+
+        Examples ::
+
+            "This is. example !" -> [["This is", "example"], [".", "!"]]
+
+        """
+        text, puncs = self._strip_to_restore(text)
+        return text, puncs
+
+    def _strip_to_restore(self, text):
+        """Auxiliary method for Punctuation.preserve()"""
+        matches = list(re.finditer(self.puncs_regular_exp, text))
+        if not matches:
+            return [text], []
+        # the text is only punctuations
+        if len(matches) == 1 and matches[0].group() == text:
+            return [], [_PUNC_IDX(text, PuncPosition.ALONE)]
+        # build a punctuation map to be used later to restore punctuations
+        puncs = []
+        for match in matches:
+            position = PuncPosition.MIDDLE
+            if match == matches[0] and text.startswith(match.group()):
+                position = PuncPosition.BEGIN
+            elif match == matches[-1] and text.endswith(match.group()):
+                position = PuncPosition.END
+            puncs.append(_PUNC_IDX(match.group(), position))
+        # convert str text to a List[str], each item is separated by a punctuation
+        splitted_text = []
+        for idx, punc in enumerate(puncs):
+            split = text.split(punc.punc)
+            prefix, suffix = split[0], punc.punc.join(split[1:])
+            splitted_text.append(prefix)
+            # if the text does not end with a punctuation, add it to the last item
+            if idx == len(puncs) - 1 and len(suffix) > 0:
+                splitted_text.append(suffix)
+            text = suffix
+        return splitted_text, puncs
+
+    @classmethod
+    def restore(cls, text, puncs):
+        """Restore punctuation in a text.
+
+        Args:
+            text (str): The text to be processed.
+            puncs (List[str]): The list of punctuations map to be used for restoring.
+
+        Examples ::
+
+            ['This is', 'example'], ['.', '!'] -> "This is. example!"
+
+        """
+        return cls._restore(text, puncs, 0)
+
+    @classmethod
+    def _restore(cls, text, puncs, num):  # pylint: disable=too-many-return-statements
+        """Auxiliary method for Punctuation.restore()"""
+        if not puncs:
+            return text
+
+        # nothing have been phonemized, returns the puncs alone
+        if not text:
+            return ["".join(m.punc for m in puncs)]
+
+        current = puncs[0]
+
+        if current.position == PuncPosition.BEGIN:
+            return cls._restore([current.punc + text[0]] + text[1:], puncs[1:], num)
+
+        if current.position == PuncPosition.END:
+            return [text[0] + current.punc] + cls._restore(text[1:], puncs[1:], num + 1)
+
+        if current.position == PuncPosition.ALONE:
+            return [current.mark] + cls._restore(text, puncs[1:], num + 1)
+
+        # POSITION == MIDDLE
+        if len(text) == 1:  # pragma: nocover
+            # a corner case where the final part of an intermediate
+            # mark (I) has not been phonemized
+            return cls._restore([text[0] + current.punc], puncs[1:], num)
+
+        return cls._restore([text[0] + current.punc + text[1]] + text[2:], puncs[1:], num)
+
+
+# if __name__ == "__main__":
+#     punc = Punctuation()
+#     text = "This is. This is, example!"
+
+#     print(punc.strip(text))
+
+#     split_text, puncs = punc.strip_to_restore(text)
+#     print(split_text, " ---- ", puncs)
+
+#     restored_text = punc.restore(split_text, puncs)
+#     print(restored_text)