rvc/lib/utils.py

import os, sys
import ffmpeg
import numpy as np
import re
import unicodedata

import logging

logging.getLogger("fairseq").setLevel(logging.WARNING)

now_dir = os.getcwd()
sys.path.append(now_dir)


def load_audio(file, sampling_rate):
    try:
        file = file.strip(" ").strip('"').strip("\n").strip('"').strip(" ")
        out, _ = (
            ffmpeg.input(file, threads=0)
            .output("-", format="f32le", acodec="pcm_f32le", ac=1, ar=sampling_rate)
            .run(cmd=["ffmpeg", "-nostdin"], capture_stdout=True, capture_stderr=True)
        )
    except Exception as error:
        raise RuntimeError(f"Failed to load audio: {error}")

    return np.frombuffer(out, np.float32).flatten()


def format_title(title):
    formatted_title = (
        unicodedata.normalize("NFKD", title).encode("ascii", "ignore").decode("utf-8")
    )
    formatted_title = re.sub(r"[\u2500-\u257F]+", "", formatted_title)
    formatted_title = re.sub(r"[^\w\s.-]", "", formatted_title)
    formatted_title = re.sub(r"\s+", "_", formatted_title)
    return formatted_title


def load_embedding(embedder_model, custom_embedder=None):
    from fairseq import checkpoint_utils  # ez

    embedder_root = os.path.join(now_dir, "rvc", "embedders")
    embedding_list = {
        "contentvec": os.path.join(embedder_root, "contentvec_base.pt"),
        "hubert": os.path.join(embedder_root, "hubert_base.pt"),
    }

    if embedder_model == "custom":
        model_path = custom_embedder
        if not custom_embedder and os.path.exists(custom_embedder):
            print("Custom embedder not found. Using the default embedder.")
            model_path = embedding_list["hubert"]
    else:
        model_path = embedding_list[embedder_model]
        if not os.path.exists(model_path):
            print("Custom embedder not found. Using the default embedder.")
            model_path = embedding_list["hubert"]

    models = checkpoint_utils.load_model_ensemble_and_task(
        [model_path],
        suffix="",
    )

    print(f"Embedding model {embedder_model} loaded successfully.")
    return models