tts.py

import base64
from abc import ABC, abstractmethod

from gtts import gTTS
from io import BytesIO
import numpy as np


class ExpressiveModel(ABC):
    @abstractmethod
    def load(self):
        pass

    @abstractmethod
    def synthesize(self, text: str, emotion: str):
        """
        Synthesis audio with emotion
        :param text: (str)
        :param emotion: (str) neutral | happy | ...
        :return: np.array
        """
        pass


class StyleTransferModel(ABC):
    @abstractmethod
    def load(self):
        pass

    @abstractmethod
    def synthesize(self, text: str, ref_audio):
        """
        Synthesis audio with reference audio
        :param text: (str)
        :param ref_audio: (np.array)
        :return: np.array
        """
        pass


class TTSService:
    """
    Get input text (str), emotion label (str) or reference audio (np.array)
    Synthesis audio (np.array)
    Convert audio to base64
    """
    @staticmethod
    def synthesis(text: str) -> str:
        tts = gTTS(text)
        # Using in-memory handling
        audio_data = BytesIO()
        tts.write_to_fp(audio_data)
        encoded_audio = base64.b64encode(audio_data.getvalue()).decode('utf-8')
        return encoded_audio
        #
        # tts.save("output.mp3")
        # with open("output.mp3", "rb") as audio_file:
        #     audio_data = audio_file.read()
        #     encoded_audio = base64.b64encode(audio_data).decode('utf-8')
        #     return encoded_audio

    @staticmethod
    def transfer(input_text: str, ref_audio: np.array) -> str:
        # Process reference audio
        # ..
        # np.array to audio
        # tts_output_np_array = np.array([0, 1, 0, 1])
        # tts_output_bytes = tts_output_np_array.tobytes()
        # audio_data = base64.b64encode(tts_output_bytes).decode('utf-8')
        # return audio_data
        #
        # example
        tts_text = input_text
        tts = gTTS(tts_text)
        audio_data = BytesIO()
        tts.write_to_fp(audio_data)
        encoded_audio = base64.b64encode(audio_data.getvalue()).decode('utf-8')
        return encoded_audio