Merge branch 'CTranslate2' into develop

config.py

@@ -7,7 +7,6 @@ import tkinter as tk
 from tkinter import font
 from models.translation.translation_languages import translation_lang
 from models.transcription.transcription_utils import getInputDevices, getDefaultInputDevice
-from models.translation.utils import ctranslate2_weights
 from utils import generatePercentageStringsList, isUniqueStrings
 
 json_serializable_vars = {}
@@ -95,10 +94,6 @@ class Config:
     def SELECTABLE_UI_LANGUAGES_DICT(self):
         return self._SELECTABLE_UI_LANGUAGES_DICT
 
-    @property
-    def CTRANSLATE2_WEIGHTS(self):
-        return self._CTRANSLATE2_WEIGHTS
-
     @property
     def MAX_MIC_ENERGY_THRESHOLD(self):
         return self._MAX_MIC_ENERGY_THRESHOLD
@@ -731,7 +726,6 @@ class Config:
             "ko": "한국어"
             # If you want to add a new language and key, please append it here.
         }
-        self._CTRANSLATE2_WEIGHTS = ctranslate2_weights
         self._MAX_MIC_ENERGY_THRESHOLD = 2000
         self._MAX_SPEAKER_ENERGY_THRESHOLD = 4000
 
@@ -805,7 +799,7 @@ class Config:
         self._AUTH_KEYS = {
             "DeepL_API": None,
         }
-        self._WEIGHT_TYPE = "small"
+        self._WEIGHT_TYPE = "m2m100_418m"
         self._SEND_MESSAGE_FORMAT = "[message]"
         self._SEND_MESSAGE_FORMAT_WITH_T = "[message]([translation])"
         self._RECEIVED_MESSAGE_FORMAT = "[message]"

main.py

@@ -9,7 +9,7 @@ if __name__ == "__main__":
 
         from config import config
         from models.translation.utils import downloadCTranslate2Weight
-        downloadCTranslate2Weight(config.PATH_LOCAL, config.WEIGHT_TYPE, config.CTRANSLATE2_WEIGHTS, splash.updateDownloadProgress)
+        downloadCTranslate2Weight(config.PATH_LOCAL, config.WEIGHT_TYPE, splash.updateDownloadProgress)
 
         import controller
         controller.createMainWindow()

model.py

@@ -63,12 +63,11 @@ class Model:
         self.speaker_audio_recorder = None
         self.speaker_energy_recorder = None
         self.speaker_energy_plot_progressbar = None
-        self.translator = Translator(config.PATH_LOCAL, config.CTRANSLATE2_WEIGHTS[config.WEIGHT_TYPE])
+        self.translator = Translator(config.PATH_LOCAL, config.WEIGHT_TYPE)
         self.keyword_processor = KeywordProcessor()
 
-    def resetTranslator(self):
-        del self.translator
-        self.translator = Translator(config.PATH_LOCAL, config.CTRANSLATE2_WEIGHTS[config.WEIGHT_TYPE])
+    def updateTranslator(self):
+        self.translator.changeCTranslate2Model(config.PATH_LOCAL, config.WEIGHT_TYPE)
 
     def resetKeywordProcessor(self):
         del self.keyword_processor

models/translation/translation_translator.py

@@ -2,25 +2,27 @@ import os
 from deepl import Translator as deepl_Translator
 from translators import translate_text as other_web_Translator
 from .translation_languages import translation_lang
+from .utils import ctranslate2_weights
 
 import ctranslate2
 import transformers
 
-TRANSLATE_MODELS = {
-    "small": "facebook/m2m100_418M",
-    "large": "facebook/m2m100_1.2B"
-}
-
 # Translator
 class Translator():
-    def __init__(self, path, weight_config):
-        self.translator_status = {}
-        directory_name = weight_config["directory_name"]
-        tokenizer = weight_config["tokenizer"]
-        self.weight_path = os.path.join(path, "weight", directory_name)
-        self.translator = ctranslate2.Translator(self.weight_path, device="cpu", device_index=0, compute_type="int8", inter_threads=1, intra_threads=4)
-        self.tokenizer = transformers.AutoTokenizer.from_pretrained(tokenizer)
+    def __init__(self, path, model_type):
         self.deepl_client = None
+        directory_name = ctranslate2_weights[model_type]["directory_name"]
+        tokenizer = ctranslate2_weights[model_type]["tokenizer"]
+        weight_path = os.path.join(path, "weight", directory_name)
+        self.ctranslate2_translator = ctranslate2.Translator(
+            weight_path,
+            device="cpu",
+            device_index=0,
+            compute_type="int8",
+            inter_threads=1,
+            intra_threads=4
+        )
+        self.ctranslate2_tokenizer = transformers.AutoTokenizer.from_pretrained(tokenizer)
 
     def authenticationDeepLAuthKey(self, authkey):
         result = True
@@ -32,6 +34,20 @@ class Translator():
             result = False
         return result
 
+    def changeCTranslate2Model(self, path, model_type):
+        directory_name = ctranslate2_weights[model_type]["directory_name"]
+        tokenizer = ctranslate2_weights[model_type]["tokenizer"]
+        weight_path = os.path.join(path, "weight", directory_name)
+        self.ctranslate2_translator = ctranslate2.Translator(
+            weight_path,
+            device="cpu",
+            device_index=0,
+            compute_type="int8",
+            inter_threads=1,
+            intra_threads=4
+        )
+        self.ctranslate2_tokenizer = transformers.AutoTokenizer.from_pretrained(tokenizer)
+
     def translate(self, translator_name, source_language, target_language, target_country, message):
         try:
             result = ""
@@ -86,12 +102,12 @@ class Translator():
                         to_language=target_language,
                         )
                 case "CTranslate2":
-                    self.tokenizer.src_lang = source_language
-                    source = self.tokenizer.convert_ids_to_tokens(self.tokenizer.encode(message))
-                    target_prefix = [self.tokenizer.lang_code_to_token[target_language]]
-                    results = self.translator.translate_batch([source], target_prefix=[target_prefix])
+                    self.ctranslate2_tokenizer.src_lang = source_language
+                    source = self.ctranslate2_tokenizer.convert_ids_to_tokens(self.ctranslate2_tokenizer.encode(message))
+                    target_prefix = [self.ctranslate2_tokenizer.lang_code_to_token[target_language]]
+                    results = self.ctranslate2_translator.translate_batch([source], target_prefix=[target_prefix])
                     target = results[0].hypotheses[0][1:]
-                    result = self.tokenizer.decode(self.tokenizer.convert_tokens_to_ids(target))
+                    result = self.ctranslate2_tokenizer.decode(self.ctranslate2_tokenizer.convert_tokens_to_ids(target))
         except Exception:
             import traceback
             with open('error.log', 'a') as f:

models/translation/utils.py

@@ -2,13 +2,13 @@ import tempfile
 from zipfile import ZipFile
 from os import path as os_path
 from os import makedirs as os_makedirs
-from requests import get as requests_get, head as requests_head
+from requests import get as requests_get
 from tqdm import tqdm
 from typing import Callable
 import hashlib
 
 ctranslate2_weights = {
-    "small": { # M2M-100 418M-parameter model
+    "m2m100_418m": { # M2M-100 418M-parameter model
         "url": "https://bit.ly/33fM1AO",
         "directory_name": "m2m100_418m",
         "tokenizer": "facebook/m2m100_418M",
@@ -18,7 +18,7 @@ ctranslate2_weights = {
             "shared_vocabulary.txt": "bd440aa21b8ca3453fc792a0018a1f3fe68b3464aadddd4d16a4b72f73c86d8c",
         }
     },
-    "large": { # M2M-100 1.2B-parameter model
+    "m2m100_12b": { # M2M-100 1.2B-parameter model
         "url": "https://bit.ly/3GYiaed",
         "directory_name": "m2m100_12b",
         "tokenizer": "facebook/m2m100_1.2b",
@@ -39,7 +39,7 @@ def calculate_file_hash(file_path, block_size=65536):
 
     return hash_object.hexdigest()
 
-def downloadCTranslate2Weight(path, weight_type="small", ctranslate2_weights=ctranslate2_weights, func=None):
+def downloadCTranslate2Weight(path, weight_type="small", func=None):
     url = ctranslate2_weights[weight_type]["url"]
     filename = 'weight.zip'
     directory_name = 'weight'