utils / rvc.py
nevreal's picture
Update rvc.py
724b3f5 verified
raw
history blame
26.9 kB
import os
import re
import random
from scipy.io.wavfile import write
from scipy.io.wavfile import read
import numpy as np
import gradio as gr
import yt_dlp
import subprocess
from original import *
import shutil, glob, subprocess
from easyfuncs import download_from_url, CachedModels, whisperspeak, whisperspeak_on, stereo_process, sr_process
os.makedirs("dataset",exist_ok=True)
os.makedirs("audios",exist_ok=True)
model_library = CachedModels()
def roformer_separator(roformer_audio, roformer_output_format="wav", roformer_overlap="4", roformer_segment_size="256"):
files_list = []
files_list.clear()
directory = "./audios"
random_id = str(random.randint(10000, 99999))
pattern = f"{random_id}"
os.makedirs("outputs", exist_ok=True)
write(f'{random_id}.wav', roformer_audio[0], roformer_audio[1])
full_roformer_model = "model_bs_roformer_ep_317_sdr_12.9755.ckpt"
prompt = f"audio-separator {random_id}.wav --model_filename {full_roformer_model} --output_dir=./outputs --output_format={roformer_output_format} --normalization=0.9 --mdxc_overlap={roformer_overlap} --mdxc_segment_size={roformer_segment_size}"
os.system(prompt)
for file in os.listdir(directory):
if re.search(pattern, file):
files_list.append(os.path.join(directory, file))
stem1_file = files_list[0]
stem2_file = files_list[1]
return stem1_file, stem2_file
with gr.Blocks(title="🔊 Nex RVC Mobile",theme=gr.themes.Base()) as app:
gr.Markdown("# Nex RVC MOBILE GUI")
with gr.Tabs():
voice_model = gr.Dropdown(label="AI Voice", choices=sorted(names), value=lambda:sorted(names)[0] if len(sorted(names)) > 0 else '', interactive=True)
refresh_button = gr.Button("Search Again", variant="primary")
with gr.TabItem("Inference"):
with gr.Row():
spk_item = gr.Slider(
minimum=0,
maximum=2333,
step=1,
label="Speaker ID",
value=0,
visible=False,
interactive=True,
)
vc_transform0 = gr.Number(
label="Pitch",
value=0
)
but0 = gr.Button(value="Convert", variant="primary")
with gr.Row():
with gr.Column():
with gr.Tabs():
with gr.TabItem("Upload"):
dropbox = gr.File(label="Drop your audio here & hit the Reload button.")
with gr.TabItem("Record"):
record_button=gr.Microphone(label="OR Record audio.", type="filepath")
with gr.TabItem("UVR", visible=False):
with gr.Row():
roformer_audio = gr.Audio(
label = "Input Audio",
type = "numpy",
interactive = True
)
with gr.Accordion("Separation by Link", open=False):
with gr.Row():
roformer_link = gr.Textbox(
label = "Link",
placeholder = "Paste the link here",
interactive = True
)
with gr.Row():
gr.Markdown("You can paste the link to the video/audio from many sites, check the complete list [here](https://github.com/yt-dlp/yt-dlp/blob/master/supportedsites.md)")
with gr.Row():
roformer_download_button = gr.Button(
"Download!",
variant = "primary"
)
roformer_download_button.click(download_audio, [roformer_link], [roformer_audio])
with gr.Row():
roformer_button = gr.Button("Separate!", variant = "primary")
with gr.Column():
roformer_stem1 = gr.Audio(
show_download_button = True,
interactive = False,
label = "Stem 1",
type = "filepath"
)
roformer_stem2 = gr.Audio(
show_download_button = True,
interactive = False,
label = "Stem 2",
type = "filepath"
)
with gr.Row():
paths_for_files = lambda path:[os.path.abspath(os.path.join(path, f)) for f in os.listdir(path) if os.path.splitext(f)[1].lower() in ('.mp3', '.wav', '.flac', '.ogg')]
input_audio0 = gr.Dropdown(
label="Input Path",
value=paths_for_files('audios')[0] if len(paths_for_files('audios')) > 0 else '',
choices=paths_for_files('audios'), # Only show absolute paths for audio files ending in .mp3, .wav, .flac or .ogg
allow_custom_value=True
)
with gr.Row():
input_player = gr.Audio(label="Input",type="numpy",interactive=False)
input_audio0.change(
inputs=[input_audio0],
outputs=[input_player],
fn=lambda path: {"value":path,"__type__":"update"} if os.path.exists(path) else None
)
record_button.stop_recording(
fn=lambda audio:audio, #TODO save wav lambda
inputs=[record_button],
outputs=[input_audio0])
dropbox.upload(
fn=lambda audio:audio.name,
inputs=[dropbox],
outputs=[input_audio0])
roformer_button.click(roformer_separator, [roformer_audio], [roformer_stem1, roformer_stem2])
with gr.Column():
with gr.Accordion("Change Index", open=False):
file_index2 = gr.Dropdown(
label="Change Index",
choices=sorted(index_paths),
interactive=True,
value=sorted(index_paths)[0] if len(sorted(index_paths)) > 0 else ''
)
index_rate1 = gr.Slider(
minimum=0,
maximum=1,
label="Index Strength",
value=0.5,
interactive=True,
)
output_player = gr.Audio(label="Output",interactive=False)
with gr.Accordion("General Settings", open=False):
f0method0 = gr.Radio(
label="Method",
choices=["pm"],
value="pm",
interactive=False,
visible=False,
)
filter_radius0 = gr.Slider(
minimum=0,
maximum=7,
label="Breathiness Reduction (Harvest only)",
value=3,
step=1,
interactive=True,
)
resample_sr0 = gr.Slider(
minimum=0,
maximum=48000,
label="Resample",
value=0,
step=1,
interactive=True,
visible=False
)
rms_mix_rate0 = gr.Slider(
minimum=0,
maximum=1,
label="Volume Normalization",
value=0,
interactive=True,
)
protect0 = gr.Slider(
minimum=0,
maximum=0.5,
label="Breathiness Protection (0 is enabled, 0.5 is disabled)",
value=0.33,
step=0.01,
interactive=True,
)
if voice_model != None:
try: vc.get_vc(voice_model.value,protect0,protect0) #load the model immediately for faster inference
except: pass
file_index1 = gr.Textbox(
label="Index Path",
interactive=True,
visible=False#Not used here
)
refresh_button.click(
fn=change_choices,
inputs=[],
outputs=[voice_model, file_index2],
api_name="infer_refresh",
)
refresh_button.click(
fn=lambda:{"choices":paths_for_files('audios'),"__type__":"update"}, #TODO check if properly returns a sorted list of audio files in the 'audios' folder that have the extensions '.wav', '.mp3', '.ogg', or '.flac'
inputs=[],
outputs = [input_audio0],
)
refresh_button.click(
fn=lambda:{"value":paths_for_files('audios')[0],"__type__":"update"} if len(paths_for_files('audios')) > 0 else {"value":"","__type__":"update"}, #TODO check if properly returns a sorted list of audio files in the 'audios' folder that have the extensions '.wav', '.mp3', '.ogg', or '.flac'
inputs=[],
outputs = [input_audio0],
)
with gr.Row():
f0_file = gr.File(label="F0 Path", visible=False)
with gr.Row():
vc_output1 = gr.Textbox(label="Information", placeholder="Welcome!",visible=False)
but0.click(
vc.vc_single,
[
spk_item,
input_audio0,
vc_transform0,
f0_file,
f0method0,
file_index1,
file_index2,
index_rate1,
filter_radius0,
resample_sr0,
rms_mix_rate0,
protect0,
],
[vc_output1, output_player],
api_name="infer_convert",
)
voice_model.change(
fn=vc.get_vc,
inputs=[voice_model, protect0, protect0],
outputs=[spk_item, protect0, protect0, file_index2, file_index2],
api_name="infer_change_voice",
)
with gr.TabItem("Download Models"):
with gr.Row():
url_input = gr.Textbox(label="URL to model (i.e. from huggingface)", value="",placeholder="https://...", scale=6)
name_output = gr.Textbox(label="Save as (if from hf, you may leave it blank)", value="",placeholder="MyModel",scale=2)
url_download = gr.Button(value="Download Model",scale=2)
url_download.click(
inputs=[url_input,name_output],
outputs=[url_input],
fn=download_from_url,
)
with gr.Row():
model_browser = gr.Dropdown(choices=list(model_library.models.keys()),label="OR Search Models (Quality UNKNOWN)",scale=5)
download_from_browser = gr.Button(value="Get",scale=2)
download_from_browser.click(
inputs=[model_browser],
outputs=[model_browser],
fn=lambda model: download_from_url(model_library.models[model],model),
)
with gr.TabItem("Train"):
with gr.Row():
with gr.Column():
training_name = gr.Textbox(label="Name your model", value="My-Voice",placeholder="My-Voice")
np7 = gr.Slider(
minimum=0,
maximum=config.n_cpu,
step=1,
label="Number of CPU processes used to extract pitch features",
value=1,
interactive=False,
visible=False
)
sr2 = gr.Radio(
label="Sampling Rate",
choices=["40k", "32k"],
value="32k",
interactive=True,
visible=True
)
if_f0_3 = gr.Radio(
label="Will your model be used for singing? If not, you can ignore this.",
choices=[True, False],
value=True,
interactive=True,
visible=False
)
version19 = gr.Radio(
label="Version",
choices=["v1", "v2"],
value="v2",
interactive=True,
visible=False,
)
dataset_folder = gr.Textbox(
label="dataset folder", value='dataset'
)
easy_uploader = gr.Files(label="Drop your audio files here",file_types=['audio'])
easy_uploader.upload(inputs=[dataset_folder],outputs=[],fn=lambda folder:os.makedirs(folder,exist_ok=True))
easy_uploader.upload(
fn=lambda files,folder: [shutil.copy2(f.name,os.path.join(folder,os.path.split(f.name)[1])) for f in files] if folder != "" else gr.Warning('Please enter a folder name for your dataset'),
inputs=[easy_uploader, dataset_folder],
outputs=[])
gpus6 = gr.Textbox(
label="Enter the GPU numbers to use separated by -, (e.g. 0-1-2)",
value="",
interactive=True,
visible=False,
)
gpu_info9 = gr.Textbox(
label="GPU Info", value=gpu_info, visible=False
)
spk_id5 = gr.Slider(
minimum=0,
maximum=4,
step=1,
label="Speaker ID",
value=0,
interactive=True,
visible=False
)
with gr.Column():
f0method8 = gr.Radio(
label="F0 extraction method",
choices=["pm", "harvest", "dio", "rmvpe", "rmvpe_gpu"],
value="pm",
interactive=False,
visible=False
)
gpus_rmvpe = gr.Textbox(
label="GPU numbers to use separated by -, (e.g. 0-1-2)",
value="",
interactive=False,
visible=False,
)
f0method8.change(
fn=change_f0_method,
inputs=[f0method8],
outputs=[gpus_rmvpe],
)
with gr.Column():
total_epoch11 = gr.Slider(
minimum=5,
maximum=1000,
step=5,
label="Epochs (more epochs may improve quality but takes longer)",
value=100,
interactive=True,
)
with gr.Accordion(label="General Settings", open=False):
gpus16 = gr.Textbox(
label="GPUs separated by -, (e.g. 0-1-2)",
value="0",
interactive=True,
visible=False,
)
save_epoch10 = gr.Slider(
minimum=1,
maximum=50,
step=1,
label="Weight Saving Frequency",
value=25,
interactive=True,
visible=False
)
batch_size12 = gr.Slider(
minimum=1,
maximum=40,
step=1,
label="Batch Size",
value=1,
interactive=True,
visible=False
)
if_save_latest13 = gr.Radio(
label="Only save the latest model",
choices=["yes", "no"],
value="yes",
interactive=True,
visible=False
)
if_cache_gpu17 = gr.Radio(
label="If your dataset is UNDER 10 minutes, cache it to train faster",
choices=["yes", "no"],
value="no",
interactive=True,
visible=True
)
if_save_every_weights18 = gr.Radio(
label="Save small model at every save point",
choices=["yes", "no"],
value="yes",
interactive=False,
visible=False
)
with gr.Accordion(label="Change pretrains", open=False):
pretrained = lambda sr, letter: [os.path.abspath(os.path.join('assets/pretrained_v2', file)) for file in os.listdir('assets/pretrained_v2') if file.endswith('.pth') and sr in file and letter in file]
pretrained_G14 = gr.Dropdown(
label="pretrained G",
# Get a list of all pretrained G model files in assets/pretrained_v2 that end with .pth
choices = pretrained(sr2.value, 'G'),
value=pretrained(sr2.value, 'G')[0] if len(pretrained(sr2.value, 'G')) > 0 else '',
interactive=True,
visible=True
)
pretrained_D15 = gr.Dropdown(
label="pretrained D",
choices = pretrained(sr2.value, 'D'),
value= pretrained(sr2.value, 'D')[0] if len(pretrained(sr2.value, 'G')) > 0 else '',
visible=True,
interactive=True
)
but1 = gr.Button("1. Process", variant="primary")
but2 = gr.Button("2. Extract Features", variant="primary")
but4 = gr.Button("3. Train Index", variant="primary")
but3 = gr.Button("4. Train Model", variant="primary")
info3 = gr.Textbox(label="Information", value="", max_lines=10)
with gr.Row():
download_model = gr.Button('5.Download Model')
with gr.Row():
model_files = gr.Files(label='Your Model and Index file can be downloaded here:')
download_model.click(
fn=lambda name: os.listdir(f'assets/weights/{name}') + glob.glob(f'logs/{name.split(".")[0]}/added_*.index'),
inputs=[training_name],
outputs=[model_files, info3])
with gr.Row():
sr2.change(
change_sr2,
[sr2, if_f0_3, version19],
[pretrained_G14, pretrained_D15],
)
version19.change(
change_version19,
[sr2, if_f0_3, version19],
[pretrained_G14, pretrained_D15, sr2],
)
if_f0_3.change(
change_f0,
[if_f0_3, sr2, version19],
[f0method8, pretrained_G14, pretrained_D15],
)
with gr.Row():
but5 = gr.Button("1 Click Training", variant="primary", visible=False)
but1.click(
preprocess_dataset,
[dataset_folder, training_name, sr2, np7],
[info3],
api_name="train_preprocess",
)
but2.click(
extract_f0_feature,
[
gpus6,
np7,
f0method8,
if_f0_3,
training_name,
version19,
gpus_rmvpe,
],
[info3],
api_name="train_extract_f0_feature",
)
but3.click(
click_train,
[
training_name,
sr2,
if_f0_3,
spk_id5,
save_epoch10,
total_epoch11,
batch_size12,
if_save_latest13,
pretrained_G14,
pretrained_D15,
gpus16,
if_cache_gpu17,
if_save_every_weights18,
version19,
],
info3,
api_name="train_start",
)
but4.click(train_index, [training_name, version19], info3)
but5.click(
train1key,
[
training_name,
sr2,
if_f0_3,
dataset_folder,
spk_id5,
np7,
f0method8,
save_epoch10,
total_epoch11,
batch_size12,
if_save_latest13,
pretrained_G14,
pretrained_D15,
gpus16,
if_cache_gpu17,
if_save_every_weights18,
version19,
gpus_rmvpe,
],
info3,
api_name="train_start_all",
)
if config.iscolab:
app.queue(max_size=20).launch(share=True,allowed_paths=["a.png"],show_error=True)
else:
app.queue(max_size=1022).launch(
server_name="0.0.0.0",
inbrowser=not config.noautoopen,
server_port=config.listen_port,
quiet=True,
)