Mercurial > hg > plosone_underreview
changeset 34:115774aff442 branch-tests
music bounds not centered but aligned with start of segment to match the way features are computed
| author | Maria Panteli |
|---|---|
| date | Thu, 14 Sep 2017 10:16:59 +0100 |
| parents | 928d9bf9224f |
| children | c4428589b82b |
| files | scripts/load_features.py tests/test_load_features.py |
| diffstat | 2 files changed, 22 insertions(+), 16 deletions(-) [+] |
line wrap: on
line diff
--- a/scripts/load_features.py Wed Sep 13 20:04:09 2017 +0100 +++ b/scripts/load_features.py Thu Sep 14 10:16:59 2017 +0100 @@ -76,13 +76,16 @@ # all segments are speech return music_idx else: - half_win_hop = int(round(0.5 * self.win2 / float(self.hop2))) + win2_frames = np.int(np.round(self.win2sec * self.framessr2)) + #half_win_hop = int(round(0.5 * self.win2 / float(self.hop2))) music_bounds = np.where(bounds[:, 2] == 'm')[0] bounds_in_frames = np.round(np.array(bounds[:, 0], dtype=float) * sr) - duration_in_frames = np.round(np.array(bounds[:, 1], dtype=float) * sr) + duration_in_frames = np.ceil(np.array(bounds[:, 1], dtype=float) * sr) for music_bound in music_bounds: - lower_bound = np.max([0, bounds_in_frames[music_bound] - half_win_hop]) - upper_bound = bounds_in_frames[music_bound] + duration_in_frames[music_bound] - half_win_hop + #lower_bound = np.max([0, bounds_in_frames[music_bound] - half_win_hop]) + #upper_bound = bounds_in_frames[music_bound] + duration_in_frames[music_bound] - half_win_hop + lower_bound = bounds_in_frames[music_bound] + upper_bound = bounds_in_frames[music_bound] + duration_in_frames[music_bound] - win2_frames music_idx.append(np.arange(lower_bound, upper_bound, dtype=int)) if len(music_idx)>0: music_idx = np.sort(np.concatenate(music_idx)) # it should be sorted, but just in case segments overlap -- remove duplicates if segments overlap
--- a/tests/test_load_features.py Wed Sep 13 20:04:09 2017 +0100 +++ b/tests/test_load_features.py Thu Sep 14 10:16:59 2017 +0100 @@ -19,14 +19,15 @@ sr = feat_loader.framessr2 music_bounds = feat_loader.get_music_idx_from_bounds(bounds, sr=sr) # upper bound minus half window size - half_win_sec = 4.0 # assume 8-second window - music_bounds_true = np.arange(np.round(sr * (np.float(bounds[-1, 1]) - half_win_sec)), dtype=int) + #half_win_sec = 4.0 # assume 8-second window + win_sec = 8 + music_bounds_true = np.arange(np.round(sr * (np.float(bounds[-1, 1]) - win_sec)), dtype=int) assert np.array_equal(music_bounds, music_bounds_true) def test_get_music_idx_from_bounds_short_segment(): # anything less than half window size is not processed - bounds = np.array([['0', '3.8', 'm']]) + bounds = np.array([['0', '7.9', 'm']]) sr = feat_loader.framessr2 music_bounds = feat_loader.get_music_idx_from_bounds(bounds, sr=sr) music_bounds_true = np.array([]) @@ -34,8 +35,8 @@ def test_get_music_idx_from_bounds_single_frame(): - bounds = np.array([['0', '4.3', 'm']]) - sr = feat_loader.framessr2 + bounds = np.array([['0', '8.1', 'm']]) + sr = feat_loader.framessr2 music_bounds = feat_loader.get_music_idx_from_bounds(bounds, sr=sr) music_bounds_true = np.array([0]) assert np.array_equal(music_bounds, music_bounds_true) @@ -47,10 +48,11 @@ ['13.5', '5.0', 'm']]) sr = feat_loader.framessr2 music_bounds = feat_loader.get_music_idx_from_bounds(bounds, sr=sr) - half_win_sec = 4.0 # assume 8-second window - music_bounds_true = np.concatenate([np.arange(np.round(sr * (10.5 - half_win_sec)), dtype=int), - np.arange(np.round(sr * (13.5 - half_win_sec)), - np.round(sr * (18.5 - half_win_sec)), dtype=int)]) + #half_win_sec = 4.0 # assume 8-second window + win_sec = 8.0 # assume 8-second window + music_bounds_true = np.concatenate([np.arange(np.round(sr * (10.5 - win_sec)), dtype=int), + np.arange(np.round(sr * 13.5), + np.round(sr * (18.5 - win_sec)), dtype=int)]) assert np.array_equal(music_bounds, music_bounds_true) @@ -61,9 +63,10 @@ sr = feat_loader.framessr2 music_bounds = feat_loader.get_music_idx_from_bounds(bounds, sr=sr) half_win_sec = 4.0 # assume 8-second window - music_bounds_true = np.concatenate([np.arange(np.round(sr * (10.5 - half_win_sec)), dtype=int), - np.arange(np.round(sr * (11.5 - half_win_sec)), - np.round(sr * (16.5 - half_win_sec)), dtype=int)]) + win_sec = 8.0 # assume 8-second window + music_bounds_true = np.concatenate([np.arange(np.round(sr * (10.5 - win_sec)), dtype=int), + np.arange(np.round(sr * 11.5), + np.round(sr * (16.5 - win_sec)), dtype=int)]) assert np.array_equal(music_bounds, music_bounds_true)