branderson
6f5e273c6c
Separated stems keep faint bleed during instrumental passages, and crepe finds a pitch in it with enough confidence to pass the voicing mask. A word whose alignment ran into such a gap was then filled with notes for the whole gap. Zero the confidence of frames more than 25 dB below the song's 90th-percentile A-weighted loudness, relative so it holds at any mastering level.
38 lines
1.4 KiB
Python
38 lines
1.4 KiB
Python
import numpy as np
|
|
from songtrace.config import Config
|
|
from songtrace.pipeline import Ctx, SongInput
|
|
from songtrace.stages import pitch
|
|
|
|
|
|
def test_pitch_stage_writes_npz(tmp_path, monkeypatch):
|
|
work = tmp_path / "w"
|
|
work.mkdir()
|
|
times = np.arange(0, 1.0, 0.01)
|
|
f0 = np.full_like(times, 220.0)
|
|
conf = np.full_like(times, 0.9)
|
|
monkeypatch.setattr(pitch, "track_pitch", lambda p, m, d: (times, f0, conf))
|
|
ctx = Ctx(inp=SongInput(title="T", artist="A", lyrics_text="x"),
|
|
cfg=Config(whisper_url="u"), work=work, log=lambda m: None)
|
|
pitch.pitch_stage().run(ctx)
|
|
data = np.load(work / "pitch.npz")
|
|
assert np.allclose(data["f0"], 220.0)
|
|
assert data["times"].shape == data["conf"].shape
|
|
|
|
|
|
def test_mute_silence_drops_bleed_but_keeps_quiet_singing():
|
|
# 60 frames of singing around -30 dB, one quiet sung frame at -50 dB, and
|
|
# 40 frames of bleed at -60 dB: only the bleed is more than 25 dB below
|
|
# the song's loud level.
|
|
loudness = np.array([-30.0] * 60 + [-50.0] + [-60.0] * 40)
|
|
conf = np.full_like(loudness, 0.8)
|
|
pitch.mute_silence(conf, loudness)
|
|
assert (conf[:61] == 0.8).all()
|
|
assert (conf[61:] == 0.0).all()
|
|
|
|
|
|
def test_mute_silence_is_relative_to_the_songs_level():
|
|
loudness = np.array([-45.0] * 60 + [-75.0] * 40)
|
|
conf = np.full_like(loudness, 0.8)
|
|
pitch.mute_silence(conf, loudness)
|
|
assert (conf[:60] == 0.8).all() and (conf[60:] == 0.0).all()
|