File size: 3,014 Bytes
9ade843 | 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 | """Assemble subset_solo20/ (symlinks + extracted 7-ch wavs) inside the AEA-annotations staging dir."""
import json, os, shutil, sys, numpy as np
from scipy.io import wavfile
H = os.path.dirname(os.path.abspath(__file__)); sys.path.insert(0, H)
from doa import ArrayDOA
B = '/mnt/work_new/weihanxu/aea_filtered_1105'; AEA = '/mnt/data_new/AEA'
OUT = '/mnt/work_new/weihanxu/hf_dataset_aea_annot_1107/subset_solo20'
sel = json.load(open(f'{H}/selected20.json'))
os.makedirs(OUT, exist_ok=True)
def link(src, dst):
os.makedirs(os.path.dirname(dst), exist_ok=True)
if os.path.lexists(dst): os.remove(dst)
os.symlink(src, dst)
recs = sorted({c['recording'] for c in sel})
for r in recs:
d = f'{OUT}/recordings/{r}'
link(f'{AEA}/{r}/recording.vrs', f'{d}/recording.vrs')
link(f'{B}/video/{r}.mp4', f'{d}/video_456px_20fps_stereo.mp4')
for root, _, fs in os.walk(f'{AEA}/{r}'):
for f in fs:
if f == 'recording.vrs' or f.startswith('.') or f.endswith('.zip'): continue
rel = os.path.relpath(os.path.join(root, f), f'{AEA}/{r}')
link(os.path.join(root, f), f'{d}/official/{rel}')
link(f'{B}/annotations/legacy/narration/{r}_narration.json', f'{d}/legacy/narration.json')
link(f'{B}/annotations/legacy/sound/{r}_sound.json', f'{d}/legacy/sound.json')
link(f'{B}/annotations/legacy/traj_gaze/{r}.json', f'{d}/legacy/traj_gaze_20fps.json')
index = []
for c in sel:
r = c['recording']; e = f'{OUT}/examples/{c["seg_id"]}'
link(f'{B}/{c["clip"]}', f'{e}/clip.mp4')
tg = json.load(open(f'{B}/annotations/legacy/traj_gaze/{r}.json'))
t0 = float(tg[0]['tracking_timestamp']) - tg[0]['frame_index'] / 20.0
wav = f'{e}/mic_array_7ch_48k.wav'
if not os.path.exists(wav):
x = ArrayDOA(f'{AEA}/{r}/recording.vrs').audio(c['clip_start_sec'] + t0, c['clip_end_sec'] + t0)
x = x / (np.abs(x).max() + 1e-9) * 0.9
wavfile.write(wav, 48000, x.astype(np.float32))
item = {k: c[k] for k in ('seg_id', 'recording', 'goal', 'steps', 'clip_start_sec', 'clip_end_sec', 'duration_sec', 'protocol_score')}
item.update({'clip_start_device_sec': round(c['clip_start_sec'] + t0, 3), 'clip_end_device_sec': round(c['clip_end_sec'] + t0, 3),
'spatial_audio_score': c['spatial_audio_score'], 'dense_doa': c['dense_doa'],
'n_annotated_sound_events': c['n_doa_events'], 'n_annotated_offaxis_events': c['n_offaxis_doa'],
'head_yaw_range_deg': c['head_yaw_range_deg'], 'sound_events': c['sound_events'],
'files': {'clip': f'examples/{c["seg_id"]}/clip.mp4', 'mic_array_wav': f'examples/{c["seg_id"]}/mic_array_7ch_48k.wav',
'recording_dir': f'recordings/{r}/'}})
json.dump(item, open(f'{e}/meta.json', 'w'), indent=1, ensure_ascii=False)
index.append(item); print('ok', c['seg_id'], flush=True)
json.dump(index, open(f'{OUT}/index.json', 'w'), indent=1, ensure_ascii=False)
print('BUILD_DONE', len(index), len(recs))
|