hotfix

mightneed to change trasnlation engine

.editorconfig

@@ -24,3 +24,7 @@ indent_style = tab
 [*.{js,css,scss,html,xml}]
 indent_style = space
 indent_size = 2
+
+[*.{dictionary,dictation,phrases}.txt]
+indent_style = tab
+indent_size = 8

README.md

@@ -14,9 +14,17 @@ creates anki hsk decks from a list of words
 ## Installation
 
 ```console
-git clone https://github.com/resemble-ai/chatterbox
+install qwen models https://github.com/QwenLM/Qwen3-TTS
+    modelscope download --model Qwen/Qwen3-TTS-12Hz-0.6B-CustomVoice --local_dir ./Qwen3-TTS-12Hz-0.6B-CustomVoice
 git clone https://gitea.wolfang.info.ve/wolfang/anki-hsk-creator
+install cuda 13.2 https://developer.nvidia.com/cuda-downloads
+install pythorch for cuda 13.2 https://pytorch.org/get-started/locally/
 git clone https://gitea.wolfang.info.ve/wolfang/anki-hsk-creator-data
+set local .env file
+HF_TOKEN =
+DATA_FOLDER =
+GWEN_FOLDER =
+ARGOS_DEVICE_TYPE = auto
 ```
 
 ## License

pyproject.toml

@@ -30,12 +30,12 @@ dependencies = [
   "pinyin-tone-converter",
   "genanki",
   "argostranslate",
-  "chatterbox-tts",
   "torch",
-  "torchaudio",
+  "soundfile",
   "torchcodec",
   "python-dotenv",
-
+  "qwen-tts",
+  # "flash-attn"
 ]
 
 [project.optional-dependencies]

src/anki_hsk_creator/__about__.py

@@ -3,4 +3,4 @@
 # SPDX-FileCopyrightText: 2026-present Wolfang Torres <wolfang.torres@gmail.com>
 #
 # SPDX-License-Identifier: GPL-3.0-or-later
-__version__ = "0.1.3"
+__version__ = "0.2.4"

src/anki_hsk_creator/__init__.py

@@ -18,3 +18,6 @@ HF_TOKEN = os.environ.get("HF_TOKEN")
 default_path = Path.home() / "anki-hsk-creator-data"
 DATA_FOLDER = Path(os.environ.get("DATA_FOLDER", default_path))
 DATA_FOLDER.mkdir(exist_ok=True, parents=True)
+default_gwen = Path.home() / "Gwen"
+GWEN_FOLDER = Path(os.environ.get("GWEN_FOLDER", default_gwen))
+GWEN_FOLDER.mkdir(exist_ok=True, parents=True)

src/anki_hsk_creator/__main__.py

@@ -5,15 +5,36 @@ from pathlib import Path
 
 # Local
 from .api import (
+    folder_proccess,
     is_file,
     list_input_files,
     pre_process_a_dictionary_file,
+    proccess_a_completed_file,
+    process_a_dictation_file,
     process_a_dictionary_file,
     process_a_phrases_file,
     select_file,
+    select_folder,
 )
-from .constants import DICT_TYPE, LANGUAGES, PHRASES_TYPE
+from .constants import (
-from .utility import ProcessFile
+    COMPLETED_TYPE,
+    DICTATION_TYPE,
+    DICT_TYPE,
+    INPUT,
+    LANGUAGES,
+    PHRASES_TYPE,
+)
+from .utility import ProcessFile, ProcessFolder
+
+
+def cli_choose_work_type() -> str:
+    """entry point for interactive interface"""
+    option = None
+    while not option in ["s", "f", "e"]:
+        option = input(
+            "Please select the work option:\n" "s:single file, f:folder, e:exit\n"
+        )
+    return option
 
 
 def cli_select_files() -> ProcessFile:
@@ -37,6 +58,30 @@ def cli_select_files() -> ProcessFile:
     return input_file
 
 
+def cli_select_folder() -> ProcessFolder:
+    """Loops until it finds a valid input_folder"""
+    print("Select data folder:")
+    in_folder = None
+    level = Path()
+    while not in_folder:
+        files = list_input_files(level)
+        for n, file in enumerate(files):
+            print(f"{n+1} - {file}")
+        s = None
+        while not s or not s.isnumeric() or not 1 <= int(s) <= len(files):
+            s = input(f"Please select the file [1-{len(files)}]: ")
+        selected = files[int(s) - 1]
+        print(f"Selected {selected}")
+        s = None
+        while s not in ("yes", "y", "no", "n"):
+            s = input("if this the folder? (yes / no, go inside) : ")
+        if s in ("yes", "y"):
+            in_folder = selected
+        else:
+            level = selected
+    return select_folder(in_folder)
+
+
 def cli_select_dictionay_tsv() -> bool:
     """If a dictionary file is selected, ask if the user wants to proccess it"""
     s = None
@@ -63,9 +108,10 @@ def cli_select_language(languages: list = None) -> str:
     for language_id, language in avaliable_languages:
         if languages and language_id in languages:
             print(f"{language_id} - {language}")
+    print("all - All languages")
     s = None
-    while not s or s not in LANGUAGES.AvailableLanguages:
+    while not s or s not in LANGUAGES.AvailableLanguages + ("all",):
-        lan_codes = [lan_id for lan_id, lan in avaliable_languages]
+        lan_codes = [lan_id for lan_id, lan in avaliable_languages] + ["all"]
         s = input(f"Please select the language {', '.join(lan_codes)}: ")
     return s
 
@@ -73,23 +119,61 @@ def cli_select_language(languages: list = None) -> str:
 def main():
     """CLI interface for the module"""
     while True:
-        input_file = cli_select_files()
+        option = cli_choose_work_type()
-        if DICT_TYPE in input_file.input_file.suffixes:
+        if option == "e":
-            dict_selected = cli_select_dictionay_tsv()
+            break
-            if dict_selected:
+        elif option == "s":
+            input_file = cli_select_files()
+            if DICT_TYPE is input_file.file_type:
+                dict_selected = cli_select_dictionay_tsv()
+                if dict_selected:
+                    language_id = cli_select_language()
+                    print(
+                        f"pre-processing {DICT_TYPE} file {input_file} with language {language_id}"
+                    )
+                    pre_process_a_dictionary_file(input_file, language_id)
+                else:
+                    print(
+                        f"Processing {DICT_TYPE} file {input_file} with language {language_id}"
+                    )
+                    language_id = cli_select_language(
+                        input_file.available_dictionary_languages
+                    )
+                    process_a_dictionary_file(input_file, language_id)
+            elif PHRASES_TYPE is input_file.file_type:
                 language_id = cli_select_language()
-                pre_process_a_dictionary_file(input_file, language_id)
+                print(
-            else:
+                    f"processing {PHRASES_TYPE} file {input_file} with language {language_id}"
-                language_id = cli_select_language(
-                    input_file.available_dictionary_languages
                 )
-                process_a_dictionary_file(input_file, language_id)
+                process_a_phrases_file(input_file, language_id)
-        elif PHRASES_TYPE in input_file.input_file.suffixes:
+            elif DICTATION_TYPE is input_file.file_type:
+                language_id = cli_select_language()
+                print(
+                    f"processing {DICTATION_TYPE} file {input_file} with language {language_id}"
+                )
+                process_a_dictation_file(input_file, language_id)
+            elif COMPLETED_TYPE is input_file.file_type:
+                language_id = input_file.input_file.suffixes[1][1:]
+                print(
+                    f"processing {COMPLETED_TYPE} file {input_file} with language {language_id}"
+                )
+                proccess_a_completed_file(input_file, language_id)
+            else:
+                print(f"File {input_file} is not recognised")
+        elif option == "f":
             language_id = cli_select_language()
-            print(
+            langs = (
-                f"processing file {input_file.input_file} with language {language_id}"
+                LANGUAGES.AvailableLanguages if language_id == "all" else [language_id]
             )
-            process_a_phrases_file(input_file, language_id)
+            in_folder = cli_select_folder()
+            for language_id in langs:
+                print(f"Selected: {in_folder} with language {language_id}")
+                for dirpath, dirnames, filenames in tuple(
+                    in_folder.absolute_input_folder.walk()
+                )[1:]:
+                    if not dirnames and filenames:
+                        folder = select_folder(dirpath.relative_to(INPUT))
+                        folder_proccess(folder, language_id=language_id)
 
 
 if __name__ == "__main__":

src/anki_hsk_creator/anki_generation.py

@@ -12,7 +12,14 @@ from genanki import Deck, Model, Note, Package
 from pinyin_tone_converter.pinyin_tone_converter import PinyinToneConverter
 
 # Local
-from .utility import DictionaryResult, ProcessFile, TranslationResult
+from .constants import COMPLETED_TYPE, DICTATION_TYPE, DICT_TYPE, PHRASES_TYPE
+from .utility import (
+    DictionaryResult,
+    ProcessFile,
+    ProcessFolder,
+    SimpleResult,
+    TranslationResult,
+)
 
 # Constants
 
@@ -136,9 +143,86 @@ HSK_MODEL = Model(
     css=CSS,
 )
 
+SIMPLE_HSK_MODEL = Model(
+    2819647620,
+    "Simple HSK Model",
+    fields=[
+        {"name": "Meaning"},
+        {"name": "Pinyin"},
+        {"name": "Simplified"},
+        {"name": "Audio"},
+    ],
+    templates=[
+        {
+            "name": "Card 1",
+            "qfmt": (
+                "<strong>{{Pinyin}}</strong>"
+                "<br>{{Meaning}}"
+                "<br>{{Audio}}"
+                "<br>Simplified: {{type:Simplified}}"
+            ),
+            "afmt": (
+                "{{FrontSide}}<hr id='answer''><div class='simple'>{{Simplified}}</div>"
+            ),
+        },
+        {
+            "name": "Card 2",
+            "qfmt": (
+                "<div class='simple'>{{Simplified}}</div>" "<br>Pinyin: {{type:Pinyin}}"
+            ),
+            "afmt": (
+                "{{FrontSide}}<hr id='answer'><strong>{{Pinyin}}</strong>"
+                "<br>{{Meaning}}<br>{{Audio}}"
+            ),
+        },
+        {
+            "name": "Card 3",
+            "qfmt": ("{{Audio}}" "<br>Simplified: {{type:Simplified}}"),
+            "afmt": (
+                "{{FrontSide}}<hr id='answer'><strong>{{Pinyin}}</strong>"
+                "<br><div class='simple'>{{Simplified}}</div>"
+            ),
+        },
+    ],
+    css=CSS,
+)
+
 # Proccess
 
 
+def output_anki_completed(
+    process_file: ProcessFile, results: list[SimpleResult]
+) -> Path:
+    """Creates an anki file for a completed file"""
+    final_file = (
+        process_file.output_name.parent / f"{process_file.input_file.stem}.apkg"
+    )
+    deck_name = "::".join(process_file.input_file.parts[:-1] + (final_file.stem,))
+    deck = Deck(
+        random.randrange(1 << 30, 1 << 31),
+        deck_name,
+        f"Deck for {process_file.input_file.stem}, "
+        "created in https://www.wolfang.info.ve/hskankicreator/",
+    )
+    audios = []
+    for result in results:
+        note = Note(
+            model=SIMPLE_HSK_MODEL,
+            fields=[
+                result.meaning,
+                PinyinToneConverter().convert_text(result.pinyin),
+                result.character,
+                f"[sound:{result.audio_path.name}]",
+            ],
+        )
+        deck.add_note(note)
+        audios.append(result.audio_path)
+    package = Package(deck)
+    package.media_files = audios
+    package.write_to_file(final_file)
+    return final_file
+
+
 def output_anki_dictation(
     process_file: ProcessFile, results: list[DictionaryResult]
 ) -> Path:
@@ -153,7 +237,6 @@ def output_anki_dictation(
         f"Deck for {final_file.name}, "
         "created in https://www.wolfang.info.ve/hskankicreator/",
     )
-    package = Package(deck)
     audios = []
     for result in results:
         note = Note(
@@ -166,6 +249,7 @@ def output_anki_dictation(
         )
         deck.add_note(note)
         audios.append(result.audio_path)
+    package = Package(deck)
     package.media_files = audios
     package.write_to_file(final_file)
     return final_file
@@ -236,3 +320,79 @@ def output_anki_phrase(
     package.media_files = audios
     package.write_to_file(final_file)
     return final_file
+
+
+def output_anki_package(
+    process_folder: ProcessFolder, results: dict[ProcessFile, list[TranslationResult]]
+):
+    final_file = process_folder.output_name.with_suffix(".apkg")
+    decks = []
+    audios = []
+    for process_file, results in results.items():
+        deck_name = "::".join(
+            process_file.input_file.parts[:-1] + (process_file.output_name.stem,)
+        )
+        deck = Deck(
+            random.randrange(1 << 30, 1 << 31),
+            deck_name,
+            f"Deck for {process_file.input_file}, "
+            "created in https://www.wolfang.info.ve/hskankicreator/",
+        )
+        if process_file.file_type is DICTATION_TYPE:
+            for result in results:
+                note = Note(
+                    model=DICTATION_MODEL,
+                    fields=[
+                        result.translated,
+                        result.line,
+                        f"[sound:{result.audio_path.name}]",
+                    ],
+                )
+                deck.add_note(note)
+                audios.append(result.audio_path)
+        elif process_file.file_type is DICT_TYPE:
+            for result in results:
+                note = Note(
+                    model=HSK_MODEL,
+                    fields=[
+                        # "\n ".join(f"{n+1}. {m}" for n, m in enumerate(result.meanings)),
+                        result.meaning,
+                        PinyinToneConverter().convert_text(result.pinyin),
+                        result.simplified,
+                        result.traditional,
+                        f"[sound:{result.audio_path.name}]",
+                    ],
+                )
+                deck.add_note(note)
+                audios.append(result.audio_path)
+        elif process_file.file_type is PHRASES_TYPE:
+            for result in results:
+                note = Note(
+                    model=PHRASE_MODEL,
+                    fields=[
+                        result.translated,
+                        result.line,
+                        f"[sound:{result.audio_path.name}]",
+                    ],
+                )
+                deck.add_note(note)
+                audios.append(result.audio_path)
+        elif process_file.file_type is COMPLETED_TYPE:
+            for result in results:
+                note = Note(
+                    model=SIMPLE_HSK_MODEL,
+                    fields=[
+                        result.meaning,
+                        PinyinToneConverter().convert_text(result.pinyin),
+                        result.character,
+                        f"[sound:{result.audio_path.name}]",
+                    ],
+                )
+                deck.add_note(note)
+                audios.append(result.audio_path)
+        decks.append(deck)
+    if decks:
+        package = Package(decks)
+        package.media_files = audios
+        package.write_to_file(final_file)
+        return final_file

src/anki_hsk_creator/api.py

@@ -4,16 +4,20 @@ Interface for managuing and procesing files
 """
 
 # Standard Library
+import re
 from pathlib import Path
 
 # Local
 from . import DATA_FOLDER
 from .anki_generation import (
+    output_anki_completed,
     output_anki_dictation,
     output_anki_dictionary,
+    output_anki_package,
     output_anki_phrase,
 )
 from .constants import (
+    COMPLETED_TYPE,
     DICTATION_TYPE,
     DICT_TYPE,
     INPUT,
@@ -23,12 +27,14 @@ from .constants import (
     RESOURCES,
 )
 from .proccessor import (
+    completed_process,
     dictation_process,
+    dictionary_bulk_process,
     dictionary_pre_process,
     dictionary_process,
     translator_process,
 )
-from .utility import CCCEDICT, TRANS, TTS, ProcessFile
+from .utility import CCCEDICT, TRANS, TTS, ProcessFile, ProcessFolder
 
 # interface
 
@@ -80,6 +86,14 @@ def select_file(file_path: Path) -> ProcessFile:
         raise ValueError(f"{file_path} is not a file")
 
 
+def select_folder(file_path: Path) -> ProcessFile:
+    """Given a relative path from `list_input_files`, return a ProcessFile"""
+    if (INPUT / file_path).is_dir():
+        return ProcessFolder(file_path)
+    else:
+        raise ValueError(f"{file_path} is not a folder")
+
+
 def create_folder(file_path: Path) -> ProcessFile:
     """Creates a folder in a file_path"""
     input_folder = INPUT / file_path
@@ -221,7 +235,9 @@ def process_a_dictation_file(process_file: ProcessFile, language_id: str) -> Pat
     with process_file.absolute_input_file.open(
         "r", encoding="utf8", newline="\n"
     ) as file:
-        text_lines = [line.strip() for line in file.read().split("。")]
+        text = file.read().strip()
+        result = re.split(r"[！？。；]*", text)
+        text_lines = [line.strip() for line in result if line.strip()]
     results = dictation_process(text_lines, process_file)
     return output_anki_dictation(process_file, results)
 
@@ -234,6 +250,73 @@ def process_a_phrases_file(process_file: ProcessFile, language_id: str) -> Path:
     with process_file.absolute_input_file.open(
         "r", encoding="utf8", newline="\n"
     ) as file:
-        text_lines = [line.strip() for line in file.readlines()]
+        text_lines = [line.strip() for line in file.readlines() if line.strip()]
     results = translator_process(text_lines, process_file)
     return output_anki_phrase(process_file, results)
+
+
+def proccess_a_completed_file(process_file: ProcessFile, language_id: str) -> Path:
+    """Process a completed file"""
+    process_file.language_id = language_id
+    TTS.create_tts()
+    with process_file.absolute_input_file.open(
+        "r", encoding="utf8", newline="\n"
+    ) as file:
+        text_lines = [line.strip() for line in file.readlines() if line.strip()]
+    results = completed_process(text_lines, process_file)
+    return output_anki_completed(process_file, results)
+
+
+def folder_proccess(process_folder: ProcessFolder, language_id: str):
+    process_folder.language_id = language_id
+    print(f"Proccesing folder {process_folder} with language {language_id}")
+    final_file = process_folder.output_name.with_suffix(".apkg")
+    if final_file.is_file():
+        print("File already proccessed")
+        return final_file
+    TTS.create_tts()
+    TRANS.create_translator(LANGUAGES.CN, language_id)
+    CCCEDICT.create_cedict(language_id)
+    results = {}
+    for process_file in process_folder.input_files:
+        print(f"Proccessing {process_file}")
+        try:
+            if process_file.file_type is DICT_TYPE:
+                with process_file.absolute_input_file.open(
+                    "r", encoding="utf8", newline="\n"
+                ) as file:
+                    words_list = [word.strip() for word in file.readlines() if word]
+                dictionary_bulk_process(words_list, process_file)
+                results[process_file] = dictionary_process(process_file)
+            elif process_file.file_type is DICTATION_TYPE:
+                with process_file.absolute_input_file.open(
+                    "r", encoding="utf8", newline="\n"
+                ) as file:
+                    text_lines = [
+                        line.strip() for line in file.read().split("。") if line.strip()
+                    ]
+                results[process_file] = dictation_process(text_lines, process_file)
+            elif process_file.file_type is PHRASES_TYPE:
+                with process_file.absolute_input_file.open(
+                    "r", encoding="utf8", newline="\n"
+                ) as file:
+                    text_lines = [
+                        line.strip() for line in file.readlines() if line.strip()
+                    ]
+                results[process_file] = translator_process(text_lines, process_file)
+            elif process_file.file_type is COMPLETED_TYPE:
+                completed_language_id = process_file.input_file.suffixes[1][1:]
+                if completed_language_id != language_id:
+                    continue
+                process_file.language_id = completed_language_id
+                with process_file.absolute_input_file.open(
+                    "r", encoding="utf8", newline="\n"
+                ) as file:
+                    text_lines = [
+                        line.strip() for line in file.readlines() if line.strip()
+                    ]
+                results[process_file] = completed_process(text_lines, process_file)
+        except (AttributeError, AssertionError):
+            print(f"Error procesing {process_file} with language {language_id}")
+            continue
+    return output_anki_package(process_folder, results)

src/anki_hsk_creator/constants.py

@@ -4,7 +4,7 @@
 import importlib.resources
 
 # Local
-from . import DATA_FOLDER
+from . import DATA_FOLDER, GWEN_FOLDER
 
 # Resources
 CCCEDICT_PATH = importlib.resources.files("anki_hsk_creator").joinpath("cedict_ts.u8")
@@ -16,16 +16,14 @@ OUTPUT = DATA_FOLDER / "output"
 OUTPUT.mkdir(exist_ok=True, parents=True)
 RESOURCES = DATA_FOLDER / "resources"
 RESOURCES.mkdir(exist_ok=True, parents=True)
+GWEN_TTS = GWEN_FOLDER / "Qwen3-TTS-12Hz-0.6B-CustomVoice"
 
 # File Types
 PHRASES_TYPE = ".phrases"
 DICT_TYPE = ".dictionary"
 DICTATION_TYPE = ".dictation"
+COMPLETED_TYPE = ".completed"
 
-# time dalays
-WORD = 0.1
-COMMA = 0.5
-POINT = 1.0
 
 class LANGUAGES:
     """Available laguages for translation"""
@@ -37,8 +35,11 @@ class LANGUAGES:
     RU = "ru"
     TR = "tr"
     TH = "th"
-
+    JA = "ja"
-    AvailableLanguages = (EN, ES, FR, RU, TR, TH)
+    KO = "ko"
+    PB = "pb"
+    VI = "vi"
+    AvailableLanguages = (ES, EN, FR, RU, TR, TH, JA, KO, PB, VI)
     LanguageNames = {
         EN: "English",
         ES: "Spanish",
@@ -46,4 +47,8 @@ class LANGUAGES:
         RU: "Russian",
         TR: "Turkish",
         TH: "Thai",
+        JA: "Japanese",
+        KO: "Korean",
+        PB: "Portuguese (Brazil)",
+        VI: "Vietnamese",
     }

src/anki_hsk_creator/proccessor.py

@@ -2,18 +2,26 @@
 
 # Standard Library
 import csv
+import uuid
 
 # Pip
 import argostranslate.translate
-import torchaudio
+import soundfile as sf
 
 # Local
-from .constants import LANGUAGES, WORD, COMMA
+from .constants import LANGUAGES
-from .utility import CCCEDICT, TTS, DictionaryResult, ProcessFile, TranslationResult
+from .utility import (
+    CCCEDICT,
+    TTS,
+    DictionaryResult,
+    ProcessFile,
+    SimpleResult,
+    TranslationResult,
+)
 
 # Constants
 
-FIELDNAMES = ["simplified", "traditional", "pinyin", "meaning"]
+FIELDNAMES = ["n", "simplified", "traditional", "pinyin", "meaning"]
 DIALECT = "excel-tab"
 
 # Results Classes
@@ -25,16 +33,15 @@ def dictation_process(
     """Process for Dictation translation"""
     results = []
     for n, line in enumerate(text_lines):
-        line = line.strip()
+        audio_line = line.strip()
-        line = "  ".join(line.split())
+        posible_name = tuple(process_file.resources.glob(f"N{n:03n}-*.wav"))
-        audio_line = f"[pause:{WORD}s]".join(line)
+        if posible_name:
-        audio_line = audio_line.replace("，", f"，[pause:{COMMA}s]")
+            audio_path = posible_name[0]
-        audio_path = process_file.resources / f"N{n:03n}.wav"
+        else:
+            audio_path = process_file.resources / f"N{n:03n}-{uuid.uuid1()}.wav"
         if not audio_path.exists():
-            audio = TTS.MODEL.generate(
+            wavs, sr = TTS.generate(f"{audio_line}。")
-                f"{line}。", language_id=LANGUAGES.CN, **TTS.DEFAULTS
+            sf.write(audio_path, wavs[0], sr)
-            )
-            torchaudio.save(audio_path, audio, TTS.MODEL.sr)
         translated = argostranslate.translate.translate(
             line, LANGUAGES.CN, process_file.language_id
         )
@@ -50,16 +57,15 @@ def translator_process(
     """Process for phases or sentence translation"""
     results = []
     for n, line in enumerate(text_lines):
-        line = line.strip()
+        audio_line = line.strip()
-        line = "  ".join(line.split())
+        posible_name = tuple(process_file.resources.glob(f"N{n:03n}-*.wav"))
-        audio_line = "[pause:0.1s]".join(line)
+        if posible_name:
-        audio_line = audio_line.replace("，", "，[pause:0.5s]")
+            audio_path = posible_name[0]
-        audio_path = process_file.resources / f"N{n:03n}.wav"
+        else:
+            audio_path = process_file.resources / f"N{n:03n}-{uuid.uuid1()}.wav"
         if not audio_path.exists():
-            audio = TTS.MODEL.generate(
+            wavs, sr = TTS.generate(f"{audio_line}。")
-                f"{audio_line}。", language_id=LANGUAGES.CN, **TTS.DEFAULTS
+            sf.write(audio_path, wavs[0], sr)
-            )
-            torchaudio.save(audio_path, audio, TTS.MODEL.sr)
         translated = argostranslate.translate.translate(
             line, LANGUAGES.CN, process_file.language_id
         )
@@ -69,6 +75,80 @@ def translator_process(
     return results
 
 
+def completed_process(
+    text_lines: list[str], process_file: ProcessFile
+) -> list[TranslationResult]:
+    """Process a complete file, not translation necesary"""
+    results = []
+    dictionary = CCCEDICT.create_cedict()
+    for n, line in enumerate(text_lines):
+        char, meaning = line.split(maxsplit=1)
+        entries = dictionary.get(char)
+        pinying = entries[0].pinyin
+        audio_path = process_file.resources / f"{char}.wav"
+        if not audio_path.is_file():
+            wavs, sr = TTS.generate(f"{char}。")
+            sf.write(audio_path, wavs[0], sr)
+        results.append(SimpleResult(char, pinying, meaning, audio_path))
+    return results
+
+
+def dictionary_bulk_process(words_list: list[str], process_file: ProcessFile):
+    dictionary_en = CCCEDICT.create_cedict(LANGUAGES.EN)
+    dictionary = CCCEDICT.create_cedict(process_file.language_id)
+    with process_file.dictionary_resource_file.open(
+        "w", encoding="utf8", newline=""
+    ) as resource_file:
+        tsv_writer = csv.DictWriter(
+            resource_file, dialect=DIALECT, fieldnames=FIELDNAMES
+        )
+        tsv_writer.writeheader()
+        number = 1
+        for words in words_list:
+            word = words.split()[0]
+            hint = words.split(maxsplit=1)[1] if len(words.split()) > 1 else None
+            entries_en = dictionary_en.get(word)
+            entries = dictionary.get(word)
+            if entries:
+                all_meanings = [
+                    meaning for entry in entries for meaning in entry.meanings
+                ]
+                pos_meanings = [
+                    meaning
+                    for entry, entry_en in zip(entries, entries_en)
+                    for meaning, meaning_en in zip(entry.meanings, entry_en.meanings)
+                    if hint and (hint in meaning_en or hint in meaning)
+                ]
+                pos_meanings = pos_meanings or all_meanings
+                if not pos_meanings:
+                    raise ValueError(
+                        f"Not menaing found for hint {hint}, {all_meanings}"
+                    )
+                meanings_text = "\n".join(
+                    f"{n+1}: {meaning}" for n, meaning in enumerate(pos_meanings)
+                )
+                pos_entries = []
+                for meaning in pos_meanings:
+                    for ent in entries:
+                        if meaning in ent.meanings:
+                            pos_entries.append(ent)
+                entry = pos_entries[0]
+                tsv_writer.writerow(
+                    {
+                        "n": number,
+                        "simplified": entry.simplified,
+                        "traditional": entry.traditional,
+                        "pinyin": entry.pinyin,
+                        "meaning": meanings_text,
+                    }
+                )
+            else:
+                print("============================================")
+                print(f"===================>ERROR: {word} not found")
+                print("============================================")
+            number += 1
+
+
 def dictionary_pre_process(words_list: list[str], process_file: ProcessFile):
     """Pre Process dictionary files into a intermediary resources file"""
     dictionary = CCCEDICT.create_cedict(process_file.language_id)
@@ -79,36 +159,52 @@ def dictionary_pre_process(words_list: list[str], process_file: ProcessFile):
             resource_file, dialect=DIALECT, fieldnames=FIELDNAMES
         )
         tsv_writer.writeheader()
+        n = 1
         for words in words_list:
             word = words.split()[0]
-            pinyin = " ".join(words.split()[1:]) if len(words.split()) > 1 else None
+            hint = " ".join(words.split()[1:]) if len(words.split()) > 1 else None
             if entries := dictionary.get(word):
-                if pinyin is not None:
+                # if pinyin is not None:
-                    entries = list(filter(lambda x: x.pinyin == pinyin, entries))
+                #     filtered_entries = tuple(filter(lambda x: x.pinyin == pinyin, entries))
                 if len(entries) > 1:
                     print(f"\nWARNING: {word} has multiple meanings:")
-                for entry in entries:
+                pos_meanings = (
+                    [
+                        entry
+                        for entry in entries
+                        for meaning in entry.meaning
+                        if hint in meaning
+                    ]
+                    if hint
+                    else []
+                )
+                pos_entries = pos_meanings or entries
+                for entry in pos_entries:
                     for meaning in entry.meanings:
-                        tsv_writer.writerow(
+                        if pos_meanings and hint in meaning or not pos_meanings:
-                            {
+                            tsv_writer.writerow(
-                                "simplified": entry.simplified,
+                                {
-                                "traditional": entry.traditional,
+                                    "n": n,
-                                "pinyin": entry.pinyin,
+                                    "simplified": entry.simplified,
-                                "meaning": meaning,
+                                    "traditional": entry.traditional,
-                            }
+                                    "pinyin": entry.pinyin,
-                        )
+                                    "meaning": meaning,
+                                }
+                            )
             else:
                 print("============================================")
                 print(f"===================>ERROR: {word} not found")
                 print("============================================")
                 tsv_writer.writerow(
                     {
+                        "n": number,
                         "simplified": word,
                         "traditional": None,
                         "pinyin": None,
                         "meaning": None,
                     }
                 )
+            n += 1
 
 
 def dictionary_process(process_file: ProcessFile) -> list[DictionaryResult]:
@@ -121,11 +217,8 @@ def dictionary_process(process_file: ProcessFile) -> list[DictionaryResult]:
         for line in reader:
             audio_path = process_file.resources / f"{line['pinyin']}.wav"
             if not audio_path.exists():
-                audio = TTS.MODEL.generate(
+                wavs, sr = TTS.generate(f"{line['simplified']}。")
-                    f"{line['simplified']}。", language_id=LANGUAGES.CN, **TTS.DEFAULTS
+                sf.write(audio_path, wavs[0], sr)
-                )
-                torchaudio.save(audio_path, audio, TTS.MODEL.sr)
-            print(line)
             result = DictionaryResult(
                 **line, audio_path=audio_path, language_id=process_file.language_id
             )

src/anki_hsk_creator/utility.py

@@ -5,6 +5,7 @@ Static clasess and functions for general use
 """
 
 # Standard Library
+import random
 from pathlib import Path
 
 # Pip
@@ -12,10 +13,21 @@ import argostranslate.package
 import argostranslate.translate
 import torch
 from cedict_utils.cedict import CedictEntry, CedictParser
-from chatterbox.mtl_tts import ChatterboxMultilingualTTS
+from qwen_tts import Qwen3TTSModel
 
 # Local
-from .constants import CCCEDICT_PATH, INPUT, LANGUAGES, OUTPUT, RESOURCES
+from .constants import (
+    CCCEDICT_PATH,
+    COMPLETED_TYPE,
+    DICTATION_TYPE,
+    DICT_TYPE,
+    GWEN_TTS,
+    INPUT,
+    LANGUAGES,
+    OUTPUT,
+    PHRASES_TYPE,
+    RESOURCES,
+)
 
 # Static Clases
 
@@ -42,21 +54,38 @@ class TRANS:
                 TRANS.PACKAGES,
             )
         )
-        print(f"available packages {packages[:5]}")
         packages_to_install = []
+        ready = False
         for in_package in packages:
+            if in_package.from_code == from_code:
+                if in_package.to_code == to_code:
+                    # Single package between 2 languages
+                    print(
+                        f"Installing package {in_package.from_code}"
+                        f"->{in_package.to_code}"
+                    )
+                    packages_to_install.append(in_package)
+                    ready = True
+                    break
+        for in_package in packages:
+            if ready:
+                break
             if in_package.from_code == from_code:
                 for out_package in packages:
                     if out_package.to_code == to_code:
                         if in_package.to_code == out_package.from_code:
                             print(
-                                f"Check in_package {in_package.from_code} {in_package.to_code}"
+                                f"Installing in_package {in_package.from_code}"
+                                f"->{in_package.to_code}"
                             )
                             print(
-                                f"Check out_package {out_package.from_code} {out_package.to_code}"
+                                f"Installing out_package {out_package.from_code}"
+                                f"->{out_package.to_code}"
                             )
                             packages_to_install.append(in_package)
                             packages_to_install.append(out_package)
+                            ready = True
+                            break
         for package in packages_to_install:
             print(f"instaling package {package}")
             argostranslate.package.install_from_path(package.download())
@@ -88,16 +117,16 @@ class TranslatedEntry:
     @property
     def meanings(self):
         """Entry translated meaning list"""
-        for meaning in self.entry.meanings:
+        if not self._translated_meanings:
-            if self.language_id != LANGUAGES.EN:
+            for meaning in self.entry.meanings:
-                print(f"translating from {LANGUAGES.EN} to {self.language_id}")
+                if self.language_id != LANGUAGES.EN:
-                print(f"-> {meaning}")
+                    trans_meaning = argostranslate.translate.translate(
-                trans_meaning = argostranslate.translate.translate(
+                        meaning, LANGUAGES.EN, self.language_id
-                    meaning, LANGUAGES.EN, self.language_id
+                    )
-                )
+                    print(f"{meaning}-> {trans_meaning}")
-            else:
+                else:
-                trans_meaning = meaning
+                    trans_meaning = meaning
-            self._translated_meanings.append(trans_meaning)
+                self._translated_meanings.append(trans_meaning)
         return self._translated_meanings
 
 
@@ -137,7 +166,12 @@ class TTS:
 
     MODEL = None
     DEVICE = None
-    DEFAULTS = {"cfg_weight": 0.2, "exaggeration": 0.8}
+    DEFAULTS = {
+        "language": "Chinese",
+        "instruct": "语速缓慢而审慎，每个音节的语调都拿捏得恰到好处，宛如教授在指导新生。",
+        "instruct": "Speak in a slow pace and enuntiate every word, as a teacher to a learning student",
+    }
+    VOICES = ["Vivian", "Serena", "Uncle_Fu", "Dylan", "Eric"]
 
     @staticmethod
     def create_tts():
@@ -145,20 +179,86 @@ class TTS:
         if TTS.DEVICE is None:
             # Automatically detect the best available device
             if torch.cuda.is_available():
-                TTS.DEVICE = "cuda"
+                TTS.DEVICE = "cuda:0"
             elif torch.backends.mps.is_available():
                 TTS.DEVICE = "mps"
             else:
                 TTS.DEVICE = "cpu"
         if TTS.MODEL is None:
-            TTS.MODEL = ChatterboxMultilingualTTS.from_pretrained(
+            TTS.MODEL = Qwen3TTSModel.from_pretrained(
-                device=TTS.DEVICE, t3_model="v3"
+                GWEN_TTS,
+                device_map=TTS.DEVICE,
+                dtype=torch.bfloat16,
+                # attn_implementation="flash_attention_2",
             )
 
+    @staticmethod
+    def generate(text: str):
+        """Generates a Waw using the defaulst values"""
+        print(f"starting to generate {text}")
+        speaker = random.choice(TTS.VOICES)
+        audio = TTS.MODEL.generate_custom_voice(
+            text=text, **TTS.DEFAULTS, speaker=speaker
+        )
+        print(f"finish to generate {text}")
+        return audio
+
 
 # Clases
 
 
+class ProcessFolder:
+    """Class that represents a folder to processs
+
+    diferent input files has direfent process_files depending on language
+    """
+
+    def __init__(self, input_folder: Path, language_id: str = None):
+        self.input_folder = input_folder
+        self._language_id = language_id
+        self.absolute_input_folder = INPUT / self.input_folder
+        # process file type
+        self.out_folder = OUTPUT / input_folder
+        self.out_folder.mkdir(parents=True, exist_ok=True)
+        self.resources = RESOURCES / input_folder
+        self.resources.mkdir(parents=True, exist_ok=True)
+
+    @property
+    def output_name(self):
+        """Posible name for the output file, still missing the filetype"""
+        if self.language_id is None:
+            raise ValueError("Not a valid language selected")
+        return self.out_folder / f"{self.input_folder.stem}.{self.language_id}.temp"
+
+    @property
+    def input_files(self):
+        input_files = []
+        for file in self.absolute_input_folder.glob(f"*.txt"):
+            for file_type in (DICT_TYPE, PHRASES_TYPE, DICTATION_TYPE, COMPLETED_TYPE):
+                if file_type in file.suffixes:
+                    input_files.append(
+                        ProcessFile(
+                            file.relative_to(INPUT), language_id=self.language_id
+                        )
+                    )
+        return input_files
+
+    @property
+    def language_id(self):
+        """language for this trasnlation process"""
+        return self._language_id
+
+    @language_id.setter
+    def language_id(self, value):
+        self._language_id = value
+
+    def __str__(self):
+        return (
+            f"Proccess {self.input_folder}"
+            f"(out: {self.out_folder}, res: {self.resources})"
+        )
+
+
 class ProcessFile:
     """Class that represents a file to processs
 
@@ -171,9 +271,31 @@ class ProcessFile:
         # process file type
         self.out_folder = OUTPUT / input_file.parent
         self.out_folder.mkdir(parents=True, exist_ok=True)
-        resources = RESOURCES / input_file
+
-        self.resources = resources.parent / resources.stem
+    @property
-        self.resources.mkdir(parents=True, exist_ok=True)
+    def resources(self):
+        resources = RESOURCES / self.input_file
+        if self.file_type is COMPLETED_TYPE:
+            resources = (
+                resources.parent / f"{self.input_file.name.split(".")[0]}.dictionary"
+            )
+        else:
+            resources = resources.parent / resources.stem
+        resources.mkdir(parents=True, exist_ok=True)
+        return resources
+
+    @property
+    def file_type(self):
+        if DICTATION_TYPE in self.input_file.suffixes:
+            return DICTATION_TYPE
+        elif DICT_TYPE in self.input_file.suffixes:
+            return DICT_TYPE
+        elif PHRASES_TYPE in self.input_file.suffixes:
+            return PHRASES_TYPE
+        elif COMPLETED_TYPE in self.input_file.suffixes:
+            return COMPLETED_TYPE
+        else:
+            raise ValueError("File type not recognized")
 
     @property
     def absolute_input_file(self):
@@ -182,7 +304,7 @@ class ProcessFile:
 
     @property
     def language_id(self):
-        """language for this trasnlation process"""
+        """language for this trasnlation proccess"""
         return self._language_id
 
     @language_id.setter
@@ -194,6 +316,8 @@ class ProcessFile:
         """Posible name for the output file, still missing the filetype"""
         if self.language_id is None:
             raise ValueError("Not a valid language selected")
+        if self.file_type is COMPLETED_TYPE:
+            return self.out_folder / f"{self.input_file.stem}.temp"
         return self.out_folder / f"{self.input_file.stem}.{self.language_id}.temp"
 
     @property
@@ -212,6 +336,19 @@ class ProcessFile:
         """for a Dictionary file loads the avaliable proceced languages"""
         return [lan.suffixes[0][1:] for lan in self.resources.glob("dictionary.*.tsv")]
 
+    def __str__(self):
+        return f"Proccess:{self.input_file}"
+
+
+class SimpleResult:
+    """Simplied result class"""
+
+    def __init__(self, character: str, pinyin: str, meaning: str, audio_path: Path):
+        self.character = character
+        self.pinyin = pinyin
+        self.meaning = meaning
+        self.audio_path = audio_path
+
 
 class TranslationResult:
     """Result of a translated process"""
@@ -234,6 +371,7 @@ class DictionaryResult:
 
     def __init__(
         self,
+        n: str,
         language_id: str,
         simplified: str,
         traditional: str,
@@ -241,6 +379,7 @@ class DictionaryResult:
         meaning: str,
         audio_path: Path,
     ):
+        self.n = n
         self.language_id = language_id
         self.simplified = simplified
         self.traditional = traditional