audio.py

import random 
import numpy as np 
from elevenlabs import voices, generate, set_api_key, UnauthenticatedRateLimitError, save
import huggingface_hub
from huggingface_hub import Repository
import os
from huggingface_hub import HfApi
import gradio as gr
from datasets import load_dataset


DATASET_REPO_URL = "https://huggingface.co/datasets/laxsvips/audiofiles"
DATA_FILENAME = "audio.mp3"
DATA_FILE = os.path.join("data", DATA_FILENAME)

api = HfApi()

HF_TOKEN = os.environ.get("HF_TOKEN")
repo = Repository(
    local_dir="data",
    clone_from=DATASET_REPO_URL,
    use_auth_token=HF_TOKEN
)

def pad_buffer(audio):
    # Pad buffer to multiple of 2 bytes
    buffer_size = len(audio)
    element_size = np.dtype(np.int16).itemsize
    if buffer_size % element_size != 0:
        audio = audio + b'\0' * (element_size - (buffer_size % element_size))
    return audio 

def generate_voice(text):
    try:
        audio = generate(
            text, 
            voice="Arnold", 
            model="eleven_monolingual_v1"
        )
        save(audio,'data/audio.mp3')   
        # save(audio,'audio.wav') 
        # commit_url = repo.push_to_hub()
        # dataset = load_dataset("audiofolder", data_dir="./data")
        audio_dataset = Dataset.from_dict({"audio": ["data/audio.mp3"]}).cast_column("audio", Audio())
        commit_url = audio_dataset.push_to_hub("laxsvips/audiofiles")

        return commit_url
        # return_url = "failure"
        # if commit_url:
        # return_url = DATASET_REPO_URL+"/"+ DATA_FILENAME
       
        # return (return_url)
        # return (44100, np.frombuffer(pad_buffer(audio), dtype=np.int16))
    except UnauthenticatedRateLimitError as e:
        raise gr.Error("Thanks for trying out ElevenLabs TTS! You've reached the free tier limit. Please provide an API key to continue.") 
    except Exception as e:
        raise gr.Error(e)