File size: 12,645 Bytes
fb0011a | 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 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 | import os
import wave
import numpy as np
import subprocess
import tempfile
def filter_audio_list(
input_path="train_list.txt",
output_path="train_list_filtered.txt",
min_duration_sec=1.25,
max_duration_sec=10.0,
min_amplitude_threshold=0.02,
max_amplitude_threshold=0.94,
ffmpeg_path="ffmpeg",
):
"""
Filter lines in input_path according to duration and amplitude checks.
Removes duplicate paths: if a given .wav path appears more than once in the
input file, ALL occurrences are removed from the output.
"""
if not os.path.exists(input_path):
print(f"Error: Input file not found at '{input_path}'")
return
kept_lines = []
removed_count = 0
total_count = 0
print(f"Starting to process '{input_path}'...")
print(f" - Minimum duration: {min_duration_sec}s")
print(f" - Maximum duration: {max_duration_sec}s")
print(f" - Minimum amplitude: {min_amplitude_threshold}")
print(f" - Maximum amplitude: {max_amplitude_threshold}")
print(
f" - Target format: 44.1kHz, 16-bit PCM, mono "
f"(will convert using {ffmpeg_path})"
)
# Read input lines (preserve full line content for output)
with open(input_path, "r", encoding="utf-8") as f:
raw_lines = f.readlines()
# Build canonical path for each non-empty line so we can detect duplicates
base_dir = os.path.dirname(os.path.abspath(input_path)) or os.getcwd()
canon_for_line = [None] * len(raw_lines)
canon_to_indices = {}
for idx, raw in enumerate(raw_lines):
ln = raw.strip()
if not ln:
continue
audio_rel = ln.split("|")[0].strip()
# Match original behavior of prefixing "../" relative to the input file
candidate = os.path.join(base_dir, "..", audio_rel)
canonical = os.path.realpath(os.path.abspath(candidate))
canon_for_line[idx] = canonical
canon_to_indices.setdefault(canonical, []).append(idx)
# Identify duplicate canonical paths (appear more than once)
duplicates_canonical = {p for p, idxs in canon_to_indices.items() if len(idxs) > 1}
if duplicates_canonical:
print(
f" [Deduplication] Found {len(duplicates_canonical)} duplicate "
"path(s). All occurrences will be removed."
)
for p in sorted(duplicates_canonical):
print(f" - {p} (occurrences={len(canon_to_indices[p])})")
# Process lines, skipping duplicates early
for idx, raw in enumerate(raw_lines):
total_count += 1
line = raw.strip()
if not line:
continue
canonical = canon_for_line[idx]
if canonical in duplicates_canonical:
# remove duplicate occurrences (do not keep any of them)
print(f" [Removing] Duplicate entry: {line.split('|')[0].strip()}")
removed_count += 1
continue
# Use the canonical absolute path for filesystem operations
audio_path = canonical
try:
if not os.path.exists(audio_path):
print(f" [Warning] File not found, skipping: {audio_path}")
removed_count += 1
continue
# Try to open as WAV; if not readable, convert with ffmpeg
try:
wav_file = wave.open(audio_path, "rb")
except wave.Error:
print(
f" [Converting] Not a readable WAV: converting {audio_path} -> "
"44.1kHz 16-bit mono WAV"
)
tmp = tempfile.NamedTemporaryFile(delete=False, suffix=".wav")
tmp.close()
cmd = [
ffmpeg_path,
"-y",
"-i",
audio_path,
"-ar",
"44100",
"-ac",
"1",
"-acodec",
"pcm_s16le",
tmp.name,
]
proc = subprocess.run(
cmd, stdout=subprocess.PIPE, stderr=subprocess.PIPE
)
if proc.returncode != 0:
stderr = proc.stderr.decode("utf-8", errors="replace")
print(
f" [Error] ffmpeg conversion failed for {audio_path}: "
f"{stderr}"
)
removed_count += 1
try:
os.unlink(tmp.name)
except Exception:
pass
continue
try:
os.replace(tmp.name, audio_path)
except Exception as e:
print(
f" [Error] Could not replace original with converted file: "
f"{e}"
)
removed_count += 1
try:
os.unlink(tmp.name)
except Exception:
pass
continue
try:
wav_file = wave.open(audio_path, "rb")
except wave.Error as e:
print(
f" [Warning] Still cannot open after conversion: "
f"{audio_path} ({e})"
)
removed_count += 1
continue
# Read header info then close
try:
num_frames = wav_file.getnframes()
framerate = wav_file.getframerate()
sampwidth = wav_file.getsampwidth()
n_channels = wav_file.getnchannels()
finally:
try:
wav_file.close()
except Exception:
pass
# Convert to target format if needed
needs_conversion = (
framerate != 44100 or sampwidth != 2 or n_channels != 1
)
if needs_conversion:
print(
f" [Converting] {audio_path} (sr={framerate}, sw={sampwidth}, "
f"ch={n_channels}) -> 44.1kHz 16-bit mono"
)
tmp = tempfile.NamedTemporaryFile(delete=False, suffix=".wav")
tmp.close()
cmd = [
ffmpeg_path,
"-y",
"-i",
audio_path,
"-ar",
"44100",
"-ac",
"1",
"-acodec",
"pcm_s16le",
tmp.name,
]
proc = subprocess.run(
cmd, stdout=subprocess.PIPE, stderr=subprocess.PIPE
)
if proc.returncode != 0:
stderr = proc.stderr.decode("utf-8", errors="replace")
print(
f" [Error] ffmpeg conversion failed for {audio_path}: "
f"{stderr}"
)
removed_count += 1
try:
os.unlink(tmp.name)
except Exception:
pass
continue
try:
os.replace(tmp.name, audio_path)
except Exception as e:
print(
f" [Error] Could not replace original with converted file: "
f"{e}"
)
removed_count += 1
try:
os.unlink(tmp.name)
except Exception:
pass
continue
# Open the (possibly converted) WAV and perform checks
try:
with wave.open(audio_path, "rb") as wf:
num_frames = wf.getnframes()
framerate = wf.getframerate()
sampwidth = wf.getsampwidth()
n_channels = wf.getnchannels()
# Duration checks
duration = num_frames / float(framerate)
if duration < min_duration_sec:
print(
f" [Removing] Short duration: {audio_path} "
f"({duration:.2f}s)"
)
removed_count += 1
continue
if duration > max_duration_sec:
print(
f" [Removing] Too long: {audio_path} "
f"({duration:.2f}s > {max_duration_sec}s)"
)
removed_count += 1
continue
# Amplitude check (16-bit PCM)
if sampwidth == 2:
frames = wf.readframes(num_frames)
if not frames:
print(f" [Removing] Empty audio: {audio_path}")
removed_count += 1
continue
samples = np.frombuffer(frames, dtype=np.int16)
if samples.size == 0:
print(f" [Removing] Empty samples: {audio_path}")
removed_count += 1
continue
normalized_samples = samples.astype(np.float32) / 32768.0
max_abs_amplitude = float(
np.max(np.abs(normalized_samples))
)
if max_abs_amplitude > max_amplitude_threshold:
print(
f" [Removing] High amplitude: {audio_path} "
f"(Peak: {max_abs_amplitude:.2f})"
)
removed_count += 1
continue
if max_abs_amplitude < min_amplitude_threshold:
print(
f" [Removing] Low amplitude: {audio_path} "
f"(Peak: {max_abs_amplitude:.2f})"
)
removed_count += 1
continue
else:
print(
f" [Info] Skipping amplitude check for {audio_path} "
f"(sample width: {sampwidth})"
)
except wave.Error as e:
print(
f" [Warning] Could not read WAV after conversion/open: "
f"{audio_path} ({e})"
)
removed_count += 1
continue
# Passed all checks
kept_lines.append(line + "\n")
except wave.Error as e:
print(
f" [Warning] Could not read WAV file, skipping: {audio_path} ({e})"
)
removed_count += 1
except FileNotFoundError as e:
print(f" [Warning] File disappeared, skipping: {audio_path} ({e})")
removed_count += 1
except Exception as e:
print(f" [Error] An unexpected error occurred on line: {line}")
print(f" {e}")
removed_count += 1
with open(output_path, "w", encoding="utf-8") as f:
f.writelines(kept_lines)
print("\n--- Processing Complete ---")
print(f"Total entries processed: {total_count}")
print(f"Entries kept: {len(kept_lines)}")
print(f"Entries removed (duration/amplitude/errors/dupes): {removed_count}")
print(f"Filtered list saved to: '{output_path}'")
if __name__ == "__main__":
# Ensure numpy and ffmpeg are available.
filter_audio_list(
input_path="train_3_filtered.txt",
output_path="train_3_filtered_1.txt",
min_duration_sec=1.25,
max_duration_sec=10.0,
min_amplitude_threshold=0.02,
max_amplitude_threshold=0.93,
ffmpeg_path="ffmpeg",
) |