Spaces:
Runtime error
Runtime error
File size: 8,085 Bytes
05aac64 |
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 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 |
import numpy as np
class Slicer:
def __init__(self, sr, threshold = -40.0, min_length = 5000, min_interval = 300, hop_size = 20, max_sil_kept = 5000):
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):
start_idx = begin * self.hop_size
return waveform[:, start_idx:min(waveform.shape[1], end * self.hop_size)] if len(waveform.shape) > 1 else waveform[start_idx:min(waveform.shape[0], end * self.hop_size)]
def slice(self, waveform):
samples = waveform.mean(axis=0) if len(waveform.shape) > 1 else 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, clip_start = None, 0
for i, rms in enumerate(rms_list):
if rms < self.threshold:
if silence_start is None: silence_start = i
continue
if silence_start is None: continue
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
if i - silence_start <= self.max_sil_kept:
pos = rms_list[silence_start : i + 1].argmin() + silence_start
sil_tags.append((0, pos) if silence_start == 0 else (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_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((rms_list[silence_start : silence_start + self.max_sil_kept + 1].argmin() + silence_start), pos), max(pos_r, pos)))
clip_start = max(pos_r, pos)
else:
pos_r = (rms_list[i - self.max_sil_kept : i + 1].argmin() + i - self.max_sil_kept)
sil_tags.append((0, pos_r) if silence_start == 0 else ((rms_list[silence_start : silence_start + self.max_sil_kept + 1].argmin() + silence_start), pos_r))
clip_start = pos_r
silence_start = None
total_frames = rms_list.shape[0]
if (silence_start is not None and total_frames - silence_start >= self.min_interval): sil_tags.append((rms_list[silence_start : min(total_frames, silence_start + self.max_sil_kept) + 1].argmin() + silence_start, total_frames + 1))
if not sil_tags: return [waveform]
else:
chunks = []
if sil_tags[0][0] > 0: chunks.append(self._apply_slice(waveform, 0, sil_tags[0][0]))
for i in range(len(sil_tags) - 1):
chunks.append(self._apply_slice(waveform, sil_tags[i][1], sil_tags[i + 1][0]))
if sil_tags[-1][1] < total_frames: chunks.append(self._apply_slice(waveform, sil_tags[-1][1], total_frames))
return chunks
class Slicer2(Slicer):
def slice2(self, waveform):
samples = waveform.mean(axis=0) if len(waveform.shape) > 1 else waveform
if samples.shape[0] <= self.min_length: return [(waveform, 0, samples.shape[0])]
rms_list = get_rms(y=samples, frame_length=self.win_size, hop_length=self.hop_size).squeeze(0)
sil_tags = []
silence_start, clip_start = None, 0
for i, rms in enumerate(rms_list):
if rms < self.threshold:
if silence_start is None: silence_start = i
continue
if silence_start is None: continue
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
if i - silence_start <= self.max_sil_kept:
pos = rms_list[silence_start : i + 1].argmin() + silence_start
sil_tags.append((0, pos) if silence_start == 0 else (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_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((rms_list[silence_start : silence_start + self.max_sil_kept + 1].argmin() + silence_start), pos), max(pos_r, pos)))
clip_start = max(pos_r, pos)
else:
pos_r = (rms_list[i - self.max_sil_kept : i + 1].argmin() + i - self.max_sil_kept)
sil_tags.append((0, pos_r) if silence_start == 0 else ((rms_list[silence_start : silence_start + self.max_sil_kept + 1].argmin() + silence_start), pos_r))
clip_start = pos_r
silence_start = None
total_frames = rms_list.shape[0]
if (silence_start is not None and total_frames - silence_start >= self.min_interval): sil_tags.append((rms_list[silence_start : min(total_frames, silence_start + self.max_sil_kept) + 1].argmin() + silence_start, total_frames + 1))
if not sil_tags: return [(waveform, 0, samples.shape[-1])]
else:
chunks = []
if sil_tags[0][0] > 0: chunks.append((self._apply_slice(waveform, 0, sil_tags[0][0]), 0, 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]), sil_tags[i][1] * self.hop_size, 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), sil_tags[-1][1] * self.hop_size, samples.shape[-1]))
return chunks
def get_rms(y, frame_length=2048, hop_length=512, pad_mode="constant"):
y = np.pad(y, (int(frame_length // 2), int(frame_length // 2)), mode=pad_mode)
axis = -1
x_shape_trimmed = list(y.shape)
x_shape_trimmed[axis] -= frame_length - 1
xw = np.moveaxis(np.lib.stride_tricks.as_strided(y, shape=tuple(x_shape_trimmed) + tuple([frame_length]), strides=y.strides + tuple([y.strides[axis]])), -1, axis - 1 if axis < 0 else axis + 1)
slices = [slice(None)] * xw.ndim
slices[axis] = slice(0, None, hop_length)
return np.sqrt(np.mean(np.abs(xw[tuple(slices)]) ** 2, axis=-2, keepdims=True))
def cut(audio, sr, db_thresh=-60, min_interval=250):
slicer = Slicer2(sr=sr, threshold=db_thresh, min_interval=min_interval)
return slicer.slice2(audio)
def restore(segments, total_len, dtype=np.float32):
out = []
last_end = 0
for start, end, processed_seg in segments:
if start > last_end: out.append(np.zeros(start - last_end, dtype=dtype))
out.append(processed_seg)
last_end = end
if last_end < total_len: out.append(np.zeros(total_len - last_end, dtype=dtype))
return np.concatenate(out, axis=-1) |