открытие сырой версии

functions/convert.py

@@ -0,0 +1,43 @@
+import subprocess
+import speech_recognition as sr
+import os
+from functions.loging import loging
+
+
+class Converter:
+    def __init__(self, path_to_file: str, message, language: str = "ru-RU"):
+        self.language = language
+        self.path_to_file = path_to_file
+        self.service = None
+        self.message = message
+
+    def audio_to_text(self) -> str:
+        import mutagen
+        f = mutagen.File(self.path_to_file)
+        if float(f.info.length) >= 30.0:
+            self.service = "google"
+            return self.google()
+        else:
+            self.service = "yandex"
+            return self.yandex()
+
+    def google(self) -> str:
+        subprocess.run(['ffmpeg', '-v', 'quiet', '-i', self.path_to_file, self.path_to_file.replace(".ogg", ".wav")])
+        os.remove(self.path_to_file)
+        wav_file = self.path_to_file.replace(".ogg", ".wav")
+        r = sr.Recognizer()
+        with sr.AudioFile(wav_file) as source:
+            audio = r.record(source)
+            r.adjust_for_ambient_noise(source)
+        response = "google\n\n" + r.recognize_google(audio, language=self.language)
+        loging(self.message, "google", wav_file, response)
+        os.remove(wav_file)
+        return response
+
+    def yandex(self) -> str:
+        from speechkit import ShortAudioRecognition, Session
+        reg = ShortAudioRecognition(Session.from_api_key("AQVN3xNJamAFP4_FS6Gis0Uud0vONFk24umBSXvh"))
+        response = "yandex\n\n" + reg.recognize(open(str(self.path_to_file), str('rb')).read())
+        loging(self.message, "yandex", self.path_to_file, response)
+        os.remove(self.path_to_file)
+        return response

functions/get_audio_messages_func.py

@@ -0,0 +1,26 @@
+from functions.convert import Converter
+from telebot import types
+
+
+def get_audio_messages_func(message: types.Message, logger, bot):
+    # достаю file_id из разнообразных полей
+    if message.content_type in ['voice']:
+        file_id = message.voice.file_id
+    elif message.content_type in ['video']:
+        file_id = message.video.file_id
+    else:
+        file_id = message.video_note.file_id
+    # создаю file_name
+    file_name = "config/" + str(message.message_id) + '.ogg'
+    # достаю имя человека
+    name = message.chat.first_name if message.chat.first_name else 'No_name'
+    # записываю в файл
+    with open(file_name, 'wb') as new_file:
+        new_file.write(bot.download_file(bot.get_file(file_id).file_path))
+    # запуск конвертора
+    converter = Converter(file_name, message)
+    message_text = converter.audio_to_text()
+    # записываю всё в logger
+    logger.info(f"Чат {name} (ID: {message.chat.id}) обработал файл {file_name}, сервисом {converter.service}")
+    # отправляю сообщение
+    bot.send_message(message.chat.id, message_text, reply_to_message_id=message.message_id)

functions/loging.py

@@ -0,0 +1,2 @@
+def loging(message, service, file_path, text):
+    print(message)