Delete audio-chatbot

audio-chatbot/app.py

@@ -1,190 +0,0 @@
-from base64 import b64encode
-from io import BytesIO
-
-from gradio import Audio, Interface, Textbox
-from gtts import gTTS
-from mtranslate import translate
-from speech_recognition import AudioFile, Recognizer
-from transformers import (BlenderbotSmallForConditionalGeneration,
-                          BlenderbotSmallTokenizer)
-
-
-# Speech to text function
-def stt(audio: object, language: str) -> str:
-    """Converts speech to text.
-    Args:
-        audio: record of user speech
-    Returns:
-        text (str): recognized speech of user
-    """
-    
-    # Create a Recognizer object
-    r = Recognizer()
-    # Open the audio file
-    with AudioFile(audio) as source:
-        # Listen for the data (load audio to memory)
-        audio_data = r.record(source)
-        # Transcribe the audio using Google's speech-to-text API
-        text = r.recognize_google(audio_data, language=language)
-    return text
-
-def to_en_translation(text: str, language: str) -> str:
-    """Translates text from specified language to English.
-    Args:
-        text (str): input text
-        language (str): desired language
-    Returns:
-        str: translated text
-    """
-    return translate(text, "en", language)
-
-
-# translating to english
-def from_en_translation(text: str, language: str) -> str:
-    """Translates text from english to specified language.
-    Args:
-        text (str): input text
-        language (str): desired language
-    Returns:
-        str: translated text
-    """
-    return translate(text, language, "en")
-
-
-class TextGenerationPipeline:
-    """Pipeline for text generation of blenderbot model.
-    Returns:
-        str: generated text
-    """
-
-    # load tokenizer and the model
-    model_name = "facebook/blenderbot_small-90M"
-    tokenizer = BlenderbotSmallTokenizer.from_pretrained(model_name)
-    model = BlenderbotSmallForConditionalGeneration.from_pretrained(model_name)
-
-    def __init__(self, **kwargs):
-        """Specififying text generation parameters.
-        For example: max_length=100 which generates text shorter than
-        100 tokens. Visit:
-        https://huggingface.co/docs/transformers/main_classes/text_generation
-        for more parameters
-        """
-        self.__dict__.update(kwargs)
-
-    def preprocess(self, text) -> str:
-        """Tokenizes input text.
-        Args:
-            text (str): user specified text
-        Returns:
-            torch.Tensor (obj): text representation as tensors
-        """
-        return self.tokenizer(text, return_tensors="pt")
-
-    def postprocess(self, outputs) -> str:
-        """Converts tensors into text.
-        Args:
-            outputs (torch.Tensor obj): model text generation output
-        Returns:
-            str: generated text
-        """
-        return self.tokenizer.decode(outputs[0], skip_special_tokens=True)
-
-    def __call__(self, text: str) -> str:
-        """Generates text from input text.
-        Args:
-            text (str): user specified text
-        Returns:
-            str: generated text
-        """
-        tokenized_text = self.preprocess(text)
-        output = self.model.generate(**tokenized_text, **self.__dict__)
-        return self.postprocess(output)
-    
-
-    
-def tts(text: str, language: str) -> object:
-    """Converts text into audio object.
-    Args:
-        text (str): generated answer of bot
-    Returns:
-        object: text to speech object
-    """
-    return gTTS(text=text, lang=language, slow=False)
-
-def tts_to_bytesio(tts_object: object) -> bytes:
-    """Converts tts object to bytes.
-    Args:
-        tts_object (object): audio object obtained from gtts
-    Returns:
-        bytes: audio bytes
-    """
-    bytes_object = BytesIO()
-    tts_object.write_to_fp(bytes_object)
-    bytes_object.seek(0)
-    return bytes_object.getvalue()
-
-
-def html_audio_autoplay(bytes: bytes) -> object:
-    """Creates html object for autoplaying audio at gradio app.
-    Args:
-        bytes (bytes): audio bytes
-    Returns:
-        object: html object that provides audio autoplaying
-    """
-    b64 = b64encode(bytes).decode()
-    html = f"""
-    <audio controls autoplay>
-    <source src="data:audio/wav;base64,{b64}" type="audio/wav">
-    </audio>
-    """
-    return html
-
-max_answer_length=100
-desired_language = "de"
-response_generator_pipe = TextGenerationPipeline(max_length=max_answer_length)
-
-
-def main(audio: object):
-    """Calls functions for deploying gradio app.
-    
-    It responds both verbally and in text
-    by taking voice input from user.
-    
-    Args:
-        audio (object): recorded speech of user
-        
-    Returns:
-        tuple containing
-        
-        - user_speech_text (str) : recognized speech
-        - bot_response_de (str) : translated answer of bot
-        - bot_response_en (str) : bot's original answer
-        - html (object) : autoplayer for bot's speech
-    """
-    user_speech_text = stt(audio, desired_language)
-    tranlated_text = to_en_translation(user_speech_text, desired_language)
-    bot_response_en = response_generator_pipe(tranlated_text)
-    bot_response_de = from_en_translation(bot_response_en, desired_language)
-    bot_voice = tts(bot_response_de, desired_language)
-    bot_voice_bytes = tts_to_bytesio(bot_voice)
-    html = html_audio_autoplay(bot_voice_bytes)
-    return user_speech_text, bot_response_de, bot_response_en, html
-
-
-Interface(
-    fn=main,
-    inputs=[
-        Audio(
-            source="microphone",
-            type="filepath",
-        ),
-    ],
-    outputs=[
-        Textbox(label="You said: "),
-        Textbox(label="AI said: "),
-        Textbox(label="AI said (English): "),
-        "html",
-    ],
-    live=True,
-    allow_flagging="never",
-).launch()

audio-chatbot/requirements.txt

@@ -1,7 +0,0 @@
-transformers==4.25.1
---find-links https://download.pytorch.org/whl/torch_stable.html
-torch==1.13.1+cpu
-gradio==3.14.0
-SpeechRecognition==3.9.0
-mtranslate==1.8
-gTTS==2.3.0