blackteay/GPT-SoVITS
1
0
Fork 0
mirror of https://github.com/RVC-Boss/GPT-SoVITS.git synced 2025-04-05 04:22:46 +08:00
Code Issues Packages Projects Releases Wiki Activity

Add files via upload

Browse Source
This commit is contained in:
RVC-Boss 2024-01-16 17:31:14 +08:00 committed by GitHub
parent e3ec41160a
commit 9e5b22eaec
No known key found for this signature in database
GPG Key ID: 4AEE18F83AFDEB23
4 changed files with 822 additions and 0 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

21
tools/my_utils.py Normal file
View File

@ -0,0 +1,21 @@
import ffmpeg
import numpy as np
def load_audio(file, sr):
try:
# https://github.com/openai/whisper/blob/main/whisper/audio.py#L26
# This launches a subprocess to decode audio while down-mixing and resampling as necessary.
# Requires the ffmpeg CLI and `ffmpeg-python` package to be installed.
file = (
file.strip(" ").strip('"').strip("\n").strip('"').strip(" ")
) # 防止小白拷路径头尾带了空格和"和回车
out, _ = (
ffmpeg.input(file, threads=0)
.output("-", format="f32le", acodec="pcm_f32le", ac=1, ar=sr)
.run(cmd=["ffmpeg", "-nostdin"], capture_stdout=True, capture_stderr=True)
)
except Exception as e:
raise RuntimeError(f"Failed to load audio: {e}")
return np.frombuffer(out, np.float32).flatten()

48
tools/slice_audio.py Normal file
View File

@ -0,0 +1,48 @@
import os,sys,numpy as np
import traceback
from scipy.io import wavfile
# parent_directory = os.path.dirname(os.path.abspath(__file__))
# sys.path.append(parent_directory)
from my_utils import load_audio
from slicer2 import Slicer
def slice(inp,opt_root,threshold,min_length,min_interval,hop_size,max_sil_kept,_max,alpha,i_part,all_part):
os.makedirs(opt_root,exist_ok=True)
if os.path.isfile(inp):
input=[inp]
elif os.path.isdir(inp):
input=["%s/%s"%(inp,name)for name in sorted(list(os.listdir(inp)))]
else:
return "输入路径存在但既不是文件也不是文件夹"
slicer = Slicer(
sr=32000, # 长音频采样率
threshold= int(threshold), # 音量小于这个值视作静音的备选切割点
min_length= int(min_length), # 每段最小多长，如果第一段太短一直和后面段连起来直到超过这个值
min_interval= int(min_interval), # 最短切割间隔
hop_size= int(hop_size), # 怎么算音量曲线，越小精度越大计算量越高（不是精度越大效果越好）
max_sil_kept= int(max_sil_kept), # 切完后静音最多留多长
)
_max=float(_max)
alpha=float(alpha)
for inp_path in input[int(i_part)::int(all_part)]:
# print(inp_path)
try:
name = os.path.basename(inp_path)
audio = load_audio(inp_path, 32000)
# print(audio.shape)
for chunk, start, end in slicer.slice(audio): # start和end是帧数
tmp_max = np.abs(chunk).max()
if(tmp_max>1):chunk/=tmp_max
chunk = (chunk / tmp_max * (_max * alpha)) + (1 - alpha) * chunk
wavfile.write(
"%s/%s_%s_%s.wav" % (opt_root, name, start, end),
32000,
# chunk.astype(np.float32),
(chunk * 32767).astype(np.int16),
)
except:
print(inp_path,"->fail->",traceback.format_exc())
return "执行完毕，请检查输出文件"
print(slice(*sys.argv[1:]))

261
tools/slicer2.py Normal file
View File

@ -0,0 +1,261 @@
import numpy as np
# This function is obtained from librosa.
def get_rms(
y,
frame_length=2048,
hop_length=512,
pad_mode="constant",
):
padding = (int(frame_length // 2), int(frame_length // 2))
y = np.pad(y, padding, mode=pad_mode)
axis = -1
# put our new within-frame axis at the end for now
out_strides = y.strides + tuple([y.strides[axis]])
# Reduce the shape on the framing axis
x_shape_trimmed = list(y.shape)
x_shape_trimmed[axis] -= frame_length - 1
out_shape = tuple(x_shape_trimmed) + tuple([frame_length])
xw = np.lib.stride_tricks.as_strided(y, shape=out_shape, strides=out_strides)
if axis < 0:
target_axis = axis - 1
else:
target_axis = axis + 1
xw = np.moveaxis(xw, -1, target_axis)
# Downsample along the target axis
slices = [slice(None)] * xw.ndim
slices[axis] = slice(0, None, hop_length)
x = xw[tuple(slices)]
# Calculate power
power = np.mean(np.abs(x) ** 2, axis=-2, keepdims=True)
return np.sqrt(power)
class Slicer:
def __init__(
self,
sr: int,
threshold: float = -40.0,
min_length: int = 5000,
min_interval: int = 300,
hop_size: int = 20,
max_sil_kept: int = 5000,
):
if not min_length >= min_interval >= hop_size:
raise ValueError(
"The following condition must be satisfied: min_length >= min_interval >= hop_size"
)
if not max_sil_kept >= hop_size:
raise ValueError(
"The following condition must be satisfied: max_sil_kept >= hop_size"
)
min_interval = sr * min_interval / 1000
self.threshold = 10 ** (threshold / 20.0)
self.hop_size = round(sr * hop_size / 1000)
self.win_size = min(round(min_interval), 4 * self.hop_size)
self.min_length = round(sr * min_length / 1000 / self.hop_size)
self.min_interval = round(min_interval / self.hop_size)
self.max_sil_kept = round(sr * max_sil_kept / 1000 / self.hop_size)
def _apply_slice(self, waveform, begin, end):
if len(waveform.shape) > 1:
return waveform[
:, begin * self.hop_size : min(waveform.shape[1], end * self.hop_size)
]
else:
return waveform[
begin * self.hop_size : min(waveform.shape[0], end * self.hop_size)
]
# @timeit
def slice(self, waveform):
if len(waveform.shape) > 1:
samples = waveform.mean(axis=0)
else:
samples = waveform
if samples.shape[0] <= self.min_length:
return [waveform]
rms_list = get_rms(
y=samples, frame_length=self.win_size, hop_length=self.hop_size
).squeeze(0)
sil_tags = []
silence_start = None
clip_start = 0
for i, rms in enumerate(rms_list):
# Keep looping while frame is silent.
if rms < self.threshold:
# Record start of silent frames.
if silence_start is None:
silence_start = i
continue
# Keep looping while frame is not silent and silence start has not been recorded.
if silence_start is None:
continue
# Clear recorded silence start if interval is not enough or clip is too short
is_leading_silence = silence_start == 0 and i > self.max_sil_kept
need_slice_middle = (
i - silence_start >= self.min_interval
and i - clip_start >= self.min_length
)
if not is_leading_silence and not need_slice_middle:
silence_start = None
continue
# Need slicing. Record the range of silent frames to be removed.
if i - silence_start <= self.max_sil_kept:
pos = rms_list[silence_start : i + 1].argmin() + silence_start
if silence_start == 0:
sil_tags.append((0, pos))
else:
sil_tags.append((pos, pos))
clip_start = pos
elif i - silence_start <= self.max_sil_kept * 2:
pos = rms_list[
i - self.max_sil_kept : silence_start + self.max_sil_kept + 1
].argmin()
pos += i - self.max_sil_kept
pos_l = (
rms_list[
silence_start : silence_start + self.max_sil_kept + 1
].argmin()
+ silence_start
)
pos_r = (
rms_list[i - self.max_sil_kept : i + 1].argmin()
+ i
- self.max_sil_kept
)
if silence_start == 0:
sil_tags.append((0, pos_r))
clip_start = pos_r
else:
sil_tags.append((min(pos_l, pos), max(pos_r, pos)))
clip_start = max(pos_r, pos)
else:
pos_l = (
rms_list[
silence_start : silence_start + self.max_sil_kept + 1
].argmin()
+ silence_start
)
pos_r = (
rms_list[i - self.max_sil_kept : i + 1].argmin()
+ i
- self.max_sil_kept
)
if silence_start == 0:
sil_tags.append((0, pos_r))
else:
sil_tags.append((pos_l, pos_r))
clip_start = pos_r
silence_start = None
# Deal with trailing silence.
total_frames = rms_list.shape[0]
if (
silence_start is not None
and total_frames - silence_start >= self.min_interval
):
silence_end = min(total_frames, silence_start + self.max_sil_kept)
pos = rms_list[silence_start : silence_end + 1].argmin() + silence_start
sil_tags.append((pos, total_frames + 1))
# Apply and return slices.
####音频+起始时间+终止时间
if len(sil_tags) == 0:
return [[waveform,0,int(total_frames*self.hop_size)]]
else:
chunks = []
if sil_tags[0][0] > 0:
chunks.append([self._apply_slice(waveform, 0, sil_tags[0][0]),0,int(sil_tags[0][0]*self.hop_size)])
for i in range(len(sil_tags) - 1):
chunks.append(
[self._apply_slice(waveform, sil_tags[i][1], sil_tags[i + 1][0]),int(sil_tags[i][1]*self.hop_size),int(sil_tags[i + 1][0]*self.hop_size)]
)
if sil_tags[-1][1] < total_frames:
chunks.append(
[self._apply_slice(waveform, sil_tags[-1][1], total_frames),int(sil_tags[-1][1]*self.hop_size),int(total_frames*self.hop_size)]
)
return chunks
def main():
import os.path
from argparse import ArgumentParser
import librosa
import soundfile
parser = ArgumentParser()
parser.add_argument("audio", type=str, help="The audio to be sliced")
parser.add_argument(
"--out", type=str, help="Output directory of the sliced audio clips"
)
parser.add_argument(
"--db_thresh",
type=float,
required=False,
default=-40,
help="The dB threshold for silence detection",
)
parser.add_argument(
"--min_length",
type=int,
required=False,
default=5000,
help="The minimum milliseconds required for each sliced audio clip",
)
parser.add_argument(
"--min_interval",
type=int,
required=False,
default=300,
help="The minimum milliseconds for a silence part to be sliced",
)
parser.add_argument(
"--hop_size",
type=int,
required=False,
default=10,
help="Frame length in milliseconds",
)
parser.add_argument(
"--max_sil_kept",
type=int,
required=False,
default=500,
help="The maximum silence length kept around the sliced clip, presented in milliseconds",
)
args = parser.parse_args()
out = args.out
if out is None:
out = os.path.dirname(os.path.abspath(args.audio))
audio, sr = librosa.load(args.audio, sr=None, mono=False)
slicer = Slicer(
sr=sr,
threshold=args.db_thresh,
min_length=args.min_length,
min_interval=args.min_interval,
hop_size=args.hop_size,
max_sil_kept=args.max_sil_kept,
)
chunks = slicer.slice(audio)
if not os.path.exists(out):
os.makedirs(out)
for i, chunk in enumerate(chunks):
if len(chunk.shape) > 1:
chunk = chunk.T
soundfile.write(
os.path.join(
out,
f"%s_%d.wav"
% (os.path.basename(args.audio).rsplit(".", maxsplit=1)[0], i),
),
chunk,
sr,
)
if __name__ == "__main__":
main()

492
tools/subfix_webui.py Normal file
View File

@ -0,0 +1,492 @@
import argparse,os
import copy
import json
import os
import uuid
import librosa
import gradio as gr
import numpy as np
import soundfile
g_json_key_text = ""
g_json_key_path = ""
g_load_file = ""
g_load_format = ""
g_max_json_index = 0
g_index = 0
g_batch = 10
g_text_list = []
g_audio_list = []
g_checkbox_list = []
g_data_json = []
def reload_data(index, batch):
global g_index
g_index = index
global g_batch
g_batch = batch
datas = g_data_json[index:index+batch]
output = []
for d in datas:
output.append(
{
g_json_key_text: d[g_json_key_text],
g_json_key_path: d[g_json_key_path]
}
)
return output
def b_change_index(index, batch):
global g_index, g_batch
g_index, g_batch = index, batch
datas = reload_data(index, batch)
output = []
for i , _ in enumerate(datas):
output.append(
# gr.Textbox(
# label=f"Text {i+index}",
# value=_[g_json_key_text]#text
# )
{
"__type__":"update",
"label":f"Text {i+index}",
"value":_[g_json_key_text]
}
)
for _ in range(g_batch - len(datas)):
output.append(
# gr.Textbox(
# label=f"Text",
# value=""
# )
{
"__type__": "update",
"label": f"Text",
"value": ""
}
)
for _ in datas:
output.append(_[g_json_key_path])
for _ in range(g_batch - len(datas)):
output.append(None)
for _ in range(g_batch):
output.append(False)
return output
def b_next_index(index, batch):
if (index + batch) <= g_max_json_index:
return index + batch , *b_change_index(index + batch, batch)
else:
return index, *b_change_index(index, batch)
def b_previous_index(index, batch):
if (index - batch) >= 0:
return index - batch , *b_change_index(index - batch, batch)
else:
return 0, *b_change_index(0, batch)
def b_submit_change(*text_list):
global g_data_json
change = False
for i, new_text in enumerate(text_list):
if g_index + i <= g_max_json_index:
new_text = new_text.strip()+' '
if (g_data_json[g_index + i][g_json_key_text] != new_text):
g_data_json[g_index + i][g_json_key_text] = new_text
change = True
if change:
b_save_file()
return g_index, *b_change_index(g_index, g_batch)
def b_delete_audio(*checkbox_list):
global g_data_json, g_index, g_max_json_index
change = False
for i, checkbox in reversed(list(enumerate(checkbox_list))):
if g_index + i < len(g_data_json):
if (checkbox == True):
g_data_json.pop(g_index + i)
change = True
g_max_json_index = len(g_data_json)-1
if g_index > g_max_json_index:
g_index = g_max_json_index
g_index = g_index if g_index >= 0 else 0
# if change:
# b_save_file()
# return gr.Slider(value=g_index, maximum=(g_max_json_index if g_max_json_index>=0 else 0)), *b_change_index(g_index, g_batch)
return {"value":g_index,"__type__":"update","maximum":(g_max_json_index if g_max_json_index>=0 else 0)},*b_change_index(g_index, g_batch)
def b_invert_selection(*checkbox_list):
new_list = [not item if item is True else True for item in checkbox_list]
return new_list
def get_next_path(filename):
base_dir = os.path.dirname(filename)
base_name = os.path.splitext(os.path.basename(filename))[0]
for i in range(100):
new_path = os.path.join(base_dir, f"{base_name}_{str(i).zfill(2)}.wav")
if not os.path.exists(new_path) :
return new_path
return os.path.join(base_dir, f'{str(uuid.uuid4())}.wav')
def b_audio_split(audio_breakpoint, *checkbox_list):
global g_data_json , g_max_json_index
checked_index = []
for i, checkbox in enumerate(checkbox_list):
if (checkbox == True and g_index+i < len(g_data_json)):
checked_index.append(g_index + i)
if len(checked_index) == 1 :
index = checked_index[0]
audio_json = copy.deepcopy(g_data_json[index])
path = audio_json[g_json_key_path]
data, sample_rate = librosa.load(path, sr=None, mono=True)
audio_maxframe = len(data)
break_frame = int(audio_breakpoint * sample_rate)
if (break_frame >= 1 and break_frame < audio_maxframe):
audio_first = data[0:break_frame]
audio_second = data[break_frame:]
nextpath = get_next_path(path)
soundfile.write(nextpath, audio_second, sample_rate)
soundfile.write(path, audio_first, sample_rate)
g_data_json.insert(index + 1, audio_json)
g_data_json[index + 1][g_json_key_path] = nextpath
b_save_file()
g_max_json_index = len(g_data_json) - 1
# return gr.Slider(value=g_index, maximum=g_max_json_index), *b_change_index(g_index, g_batch)
return {"value":g_index,"maximum":g_max_json_index,"__type__":"update"}, *b_change_index(g_index, g_batch)
def b_merge_audio(interval_r, *checkbox_list):
global g_data_json , g_max_json_index
checked_index = []
audios_path = []
audios_text = []
for i, checkbox in enumerate(checkbox_list):
if (checkbox == True and g_index+i < len(g_data_json)):
checked_index.append(g_index + i)
if (len(checked_index)>1):
for i in checked_index:
audios_path.append(g_data_json[i][g_json_key_path])
audios_text.append(g_data_json[i][g_json_key_text])
for i in reversed(checked_index[1:]):
g_data_json.pop(i)
base_index = checked_index[0]
base_path = audios_path[0]
g_data_json[base_index][g_json_key_text] = "".join(audios_text)
audio_list = []
l_sample_rate = None
for i, path in enumerate(audios_path):
data, sample_rate = librosa.load(path, sr=l_sample_rate, mono=True)
l_sample_rate = sample_rate
if (i > 0):
silence = np.zeros(int(l_sample_rate * interval_r))
audio_list.append(silence)
audio_list.append(data)
audio_concat = np.concatenate(audio_list)
soundfile.write(base_path, audio_concat, l_sample_rate)
b_save_file()
g_max_json_index = len(g_data_json) - 1
# return gr.Slider(value=g_index, maximum=g_max_json_index), *b_change_index(g_index, g_batch)
return {"value":g_index,"maximum":g_max_json_index,"__type__":"update"}, *b_change_index(g_index, g_batch)
def b_save_json():
with open(g_load_file,'w', encoding="utf-8") as file:
for data in g_data_json:
file.write(f'{json.dumps(data, ensure_ascii = False)}\n')
def b_save_list():
with open(g_load_file,'w', encoding="utf-8") as file:
for data in g_data_json:
wav_path = data["wav_path"]
speaker_name = data["speaker_name"]
language = data["language"]
text = data["text"]
file.write(f"{wav_path}|{speaker_name}|{language}|{text}".strip()+'\n')
def b_load_json():
global g_data_json, g_max_json_index
with open(g_load_file, 'r', encoding="utf-8") as file:
g_data_json = file.readlines()
g_data_json = [json.loads(line) for line in g_data_json]
g_max_json_index = len(g_data_json) - 1
def b_load_list():
global g_data_json, g_max_json_index
with open(g_load_file, 'r', encoding="utf-8") as source:
data_list = source.readlines()
for _ in data_list:
data = _.split('|')
if (len(data) == 4):
wav_path, speaker_name, language, text = data
g_data_json.append(
{
'wav_path':wav_path,
'speaker_name':speaker_name,
'language':language,
'text':text.strip()
}
)
else:
print("error line:", data)
g_max_json_index = len(g_data_json) - 1
def b_save_file():
if g_load_format == "json":
b_save_json()
elif g_load_format == "list":
b_save_list()
def b_load_file():
if g_load_format == "json":
b_load_json()
elif g_load_format == "list":
b_load_list()
def set_global(load_json, load_list, json_key_text, json_key_path, batch):
global g_json_key_text, g_json_key_path, g_load_file, g_load_format, g_batch
g_batch = int(batch)
if (load_json != "None"):
g_load_format = "json"
g_load_file = load_json
elif (load_list != "None"):
g_load_format = "list"
g_load_file = load_list
else:
g_load_format = "list"
g_load_file = "demo.list"
g_json_key_text = json_key_text
g_json_key_path = json_key_path
b_load_file()
if __name__ == "__main__":
parser = argparse.ArgumentParser(description='Process some integers.')
parser.add_argument('--load_json', default="None", help='source file, like demo.json')
parser.add_argument('--load_list', default="None", help='source file, like demo.list')
parser.add_argument('--webui_port_subfix', default=9871, help='source file, like demo.list')
parser.add_argument('--json_key_text', default="text", help='the text key name in json, Default: text')
parser.add_argument('--json_key_path', default="wav_path", help='the path key name in json, Default: wav_path')
parser.add_argument('--g_batch', default=10, help='max number g_batch wav to display, Default: 10')
args = parser.parse_args()
set_global(args.load_json, args.load_list, args.json_key_text, args.json_key_path, args.g_batch)
with gr.Blocks() as demo:
with gr.Row():
btn_change_index = gr.Button("Change Index")
btn_submit_change = gr.Button("Submit Text")
btn_merge_audio = gr.Button("Merge Audio")
btn_delete_audio = gr.Button("Delete Audio")
btn_previous_index = gr.Button("Previous Index")
btn_next_index = gr.Button("Next Index")
with gr.Row():
index_slider = gr.Slider(
minimum=0, maximum=g_max_json_index, value=g_index, step=1, label="Index", scale=3
)
splitpoint_slider = gr.Slider(
minimum=0, maximum=120.0, value=0, step=0.1, label="Audio Split Point(s)", scale=3
)
btn_audio_split = gr.Button("Split Audio", scale=1)
btn_save_json = gr.Button("Save File", visible=True, scale=1)
btn_invert_selection = gr.Button("Invert Selection", scale=1)
with gr.Row():
with gr.Column():
for _ in range(0,g_batch):
with gr.Row():
text = gr.Textbox(
label = "Text",
visible = True,
scale=5
)
audio_output = gr.Audio(
label="Output Audio",
visible = True,
scale=5
)
audio_check = gr.Checkbox(
label="Yes",
show_label = True,
info = "Choose Audio",
scale=1
)
g_text_list.append(text)
g_audio_list.append(audio_output)
g_checkbox_list.append(audio_check)
with gr.Row():
batchsize_slider = gr.Slider(
minimum=1, maximum=g_batch, value=g_batch, step=1, label="Batch Size", scale=3, interactive=False
)
interval_slider = gr.Slider(
minimum=0, maximum=2, value=0, step=0.01, label="Interval", scale=3
)
btn_theme_dark = gr.Button("Light Theme", link="?__theme=light", scale=1)
btn_theme_light = gr.Button("Dark Theme", link="?__theme=dark", scale=1)
btn_change_index.click(
b_change_index,
inputs=[
index_slider,
batchsize_slider,
],
outputs=[
*g_text_list,
*g_audio_list,
*g_checkbox_list
],
)
btn_submit_change.click(
b_submit_change,
inputs=[
*g_text_list,
],
outputs=[
index_slider,
*g_text_list,
*g_audio_list,
*g_checkbox_list
],
)
btn_previous_index.click(
b_previous_index,
inputs=[
index_slider,
batchsize_slider,
],
outputs=[
index_slider,
*g_text_list,
*g_audio_list,
*g_checkbox_list
],
)
btn_next_index.click(
b_next_index,
inputs=[
index_slider,
batchsize_slider,
],
outputs=[
index_slider,
*g_text_list,
*g_audio_list,
*g_checkbox_list
],
)
btn_delete_audio.click(
b_delete_audio,
inputs=[
*g_checkbox_list
],
outputs=[
index_slider,
*g_text_list,
*g_audio_list,
*g_checkbox_list
]
)
btn_merge_audio.click(
b_merge_audio,
inputs=[
interval_slider,
*g_checkbox_list
],
outputs=[
index_slider,
*g_text_list,
*g_audio_list,
*g_checkbox_list
]
)
btn_audio_split.click(
b_audio_split,
inputs=[
splitpoint_slider,
*g_checkbox_list
],
outputs=[
index_slider,
*g_text_list,
*g_audio_list,
*g_checkbox_list
]
)
btn_invert_selection.click(
b_invert_selection,
inputs=[
*g_checkbox_list
],
outputs=[
*g_checkbox_list
]
)
btn_save_json.click(
b_save_file
)
demo.load(
b_change_index,
inputs=[
index_slider,
batchsize_slider,
],
outputs=[
*g_text_list,
*g_audio_list,
*g_checkbox_list
],
)
demo.launch(
server_name="0.0.0.0",
inbrowser=True,
quiet=True,
server_port=int(args.webui_port_subfix)
)