|
import io |
|
|
|
import gradio as gr |
|
import librosa |
|
import numpy as np |
|
import soundfile |
|
import torch |
|
from inference.infer_tool import Svc |
|
import logging |
|
|
|
logging.getLogger('numba').setLevel(logging.WARNING) |
|
|
|
model_name = "logs/G_14000.pth" |
|
config_name = "config/config_cris.json" |
|
|
|
svc_model = Svc(model_name, config_name) |
|
sid_map = { |
|
"cris": "cris" |
|
} |
|
|
|
|
|
def vc_fn(sid, input_audio, vc_transform): |
|
if input_audio is None: |
|
return "You need to upload an audio", None |
|
sampling_rate, audio = input_audio |
|
|
|
duration = audio.shape[0] / sampling_rate |
|
audio = (audio / np.iinfo(audio.dtype).max).astype(np.float32) |
|
if len(audio.shape) > 1: |
|
audio = librosa.to_mono(audio.transpose(1, 0)) |
|
if sampling_rate != 16000: |
|
audio = librosa.resample(audio, orig_sr=sampling_rate, target_sr=16000) |
|
print(audio.shape) |
|
out_wav_path = io.BytesIO() |
|
soundfile.write(out_wav_path, audio, 16000, format="wav") |
|
out_wav_path.seek(0) |
|
|
|
sid = sid_map[sid] |
|
out_audio, out_sr = svc_model.infer(sid, vc_transform, out_wav_path) |
|
_audio = out_audio.cpu().numpy() |
|
return "Success", (32000, _audio) |
|
|
|
|
|
app = gr.Blocks() |
|
with app: |
|
with gr.Tabs(): |
|
with gr.TabItem("Basic"): |
|
gr.Markdown(value=""" |
|
这是sovits 3.0 32khz版本澄闪语音的在线demo |
|
|
|
|
|
如果要在本地使用该demo，请使用git lfs clone 该仓库，安装requirements.txt后运行app.py即可 |
|
|
|
项目改写基于 https://huggingface.co/spaces/innnky/nyaru-svc-3.0 |
|
|
|
本地合成可以删除26、27两行代码以解除合成45s长度限制""") |
|
sid = gr.Dropdown(label="音色", choices=["cris"], value="cris") |
|
vc_input3 = gr.Audio(label="上传音频（长度小于45秒）") |
|
vc_transform = gr.Number(label="变调（整数，可以正负，半音数量，升高八度就是12，男生一般从12开始，女生可以试试从1开始微调）", value=0) |
|
vc_submit = gr.Button("转换", variant="primary") |
|
vc_output1 = gr.Textbox(label="Output Message") |
|
vc_output2 = gr.Audio(label="Output Audio") |
|
vc_submit.click(vc_fn, [sid, vc_input3, vc_transform], [vc_output1, vc_output2]) |
|
|
|
app.launch() |
|
|
