Spaces:
Running
on
Zero
Running
on
Zero
File size: 2,599 Bytes
4efe6b5 |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 |
import os, sys
import librosa
import soundfile as sf
import numpy as np
import re
import unicodedata
from fairseq import checkpoint_utils
import wget
import logging
logging.getLogger("fairseq").setLevel(logging.WARNING)
now_dir = os.getcwd()
sys.path.append(now_dir)
def load_audio(file, sample_rate):
try:
file = file.strip(" ").strip('"').strip("\n").strip('"').strip(" ")
audio, sr = sf.read(file)
if len(audio.shape) > 1:
audio = librosa.to_mono(audio.T)
if sr != sample_rate:
audio = librosa.resample(audio, orig_sr=sr, target_sr=sample_rate)
except Exception as error:
raise RuntimeError(f"An error occurred loading the audio: {error}")
return audio.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):
embedder_root = os.path.join(now_dir, "rvc", "models", "embedders")
embedding_list = {
"contentvec": os.path.join(embedder_root, "contentvec_base.pt"),
"japanese-hubert-base": os.path.join(embedder_root, "japanese-hubert-base.pt"),
"chinese-hubert-large": os.path.join(embedder_root, "chinese-hubert-large.pt"),
}
online_embedders = {
"japanese-hubert-base": "https://huggingface.co./rinna/japanese-hubert-base/resolve/main/fairseq/model.pt",
"chinese-hubert-large": "https://huggingface.co./TencentGameMate/chinese-hubert-large/resolve/main/chinese-hubert-large-fairseq-ckpt.pt",
}
if embedder_model == "custom":
model_path = custom_embedder
if not custom_embedder and os.path.exists(custom_embedder):
model_path = embedding_list["contentvec"]
else:
model_path = embedding_list[embedder_model]
if embedder_model in online_embedders:
model_path = embedding_list[embedder_model]
url = online_embedders[embedder_model]
print(f"\nDownloading {url} to {model_path}...")
wget.download(url, out=model_path)
else:
model_path = embedding_list["contentvec"]
models = checkpoint_utils.load_model_ensemble_and_task(
[model_path],
suffix="",
)
# print(f"Embedding model {embedder_model} loaded successfully.")
return models
|