Spaces:
Sleeping
Sleeping
File size: 10,186 Bytes
5b3249e aa5c284 5b3249e adb1607 77bffac 5b3249e adb1607 77bffac 5b3249e aa5c284 5b3249e adb1607 5b3249e 77bffac 5b3249e adb1607 aa5c284 adb1607 5b3249e | 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 | import unittest
from pathlib import Path
import sys
import numpy as np
sys.path.insert(0, str(Path(__file__).resolve().parents[1]))
from harmonic_pipeline import (
BassHint,
ChordCandidate,
KeyCandidate,
candidate_to_payload,
empty_harmonic_response,
build_key_candidates,
build_stage_progression,
estimate_beat_metrics,
estimate_meter_hint,
estimate_no_chord_evidence,
merge_consecutive_events,
normalize_vector,
rank_progression_for_key,
HarmonicSegment,
build_segment_chord_candidates,
)
from main import AnalisarRequest
class HarmonicPipelineTests(unittest.TestCase):
def test_no_chord_detects_digital_silence(self):
evidence = estimate_no_chord_evidence(
np.zeros(16000, dtype=np.float32),
sr=16000,
chroma=np.zeros((12, 24), dtype=np.float32),
instrumento="teclado",
)
self.assertTrue(evidence.detected)
self.assertGreaterEqual(evidence.confidence, evidence.threshold)
self.assertIn("digital_silence", evidence.reasons)
def test_no_chord_detects_diffuse_noise(self):
rng = np.random.default_rng(20260810)
noise = rng.normal(0.0, 0.08, 32000).astype(np.float32)
diffuse_chroma = np.full((12, 40), 1.0 / 12.0, dtype=np.float32)
evidence = estimate_no_chord_evidence(
noise,
sr=16000,
chroma=diffuse_chroma,
instrumento="violao",
)
self.assertTrue(evidence.detected)
self.assertIn("weak_chord_template_fit", evidence.reasons)
self.assertIn("high_chroma_entropy", evidence.reasons)
def test_no_chord_detects_speech_like_sparse_unstable_content(self):
time = np.arange(32000, dtype=np.float32) / 16000.0
speech_like = (
0.05 * np.sin(2.0 * np.pi * (145.0 + (38.0 * time)) * time)
+ 0.018 * np.sin(2.0 * np.pi * (290.0 + (71.0 * time)) * time)
).astype(np.float32)
sparse_chroma = np.zeros((12, 48), dtype=np.float32)
for frame in range(sparse_chroma.shape[1]):
sparse_chroma[frame % 12, frame] = 1.0
evidence = estimate_no_chord_evidence(
speech_like,
sr=16000,
chroma=sparse_chroma,
instrumento="teclado",
)
self.assertTrue(evidence.detected)
self.assertIn("sparse_unstable_tonal_content", evidence.reasons)
def test_no_chord_does_not_reject_sparse_melodic_instrument_phrase(self):
time = np.arange(32000, dtype=np.float32) / 16000.0
melodic_audio = (0.05 * np.sin(2.0 * np.pi * 220.0 * time)).astype(np.float32)
melodic_chroma = np.zeros((12, 48), dtype=np.float32)
phrase = [9, 11, 1, 4, 6, 9]
for frame in range(melodic_chroma.shape[1]):
melodic_chroma[phrase[(frame // 8) % len(phrase)], frame] = 1.0
evidence = estimate_no_chord_evidence(
melodic_audio,
sr=16000,
chroma=melodic_chroma,
instrumento="sax_alto",
)
self.assertFalse(evidence.detected)
def test_no_chord_preserves_quiet_tonal_chord(self):
time = np.arange(32000, dtype=np.float32) / 16000.0
quiet_c_major = (
0.00018 * np.sin(2.0 * np.pi * 261.63 * time)
+ 0.00018 * np.sin(2.0 * np.pi * 329.63 * time)
+ 0.00018 * np.sin(2.0 * np.pi * 392.00 * time)
).astype(np.float32)
chord_chroma = np.zeros((12, 40), dtype=np.float32)
chord_chroma[0, :] = 1.0
chord_chroma[4, :] = 0.95
chord_chroma[7, :] = 0.98
evidence = estimate_no_chord_evidence(
quiet_c_major,
sr=16000,
chroma=chord_chroma,
instrumento="teclado",
)
self.assertFalse(evidence.detected)
self.assertGreater(evidence.template_fit, 0.9)
def test_empty_response_exposes_no_chord_without_changing_legacy_shapes(self):
evidence = estimate_no_chord_evidence(
np.zeros(16000, dtype=np.float32),
sr=16000,
chroma=np.zeros((12, 8), dtype=np.float32),
instrumento="ukulele",
)
response = empty_harmonic_response(no_chord_evidence=evidence)
self.assertEqual(response["acorde_atual"], "")
self.assertEqual(response["acordes"], [])
self.assertEqual(response["current_chord_candidates"], [])
diagnostic = response["diagnostico_harmonico"]
self.assertTrue(diagnostic["no_chord_detected"])
self.assertEqual(diagnostic["silence_ratio"], 1.0)
self.assertEqual(diagnostic["score_kind"], "heuristic_evidence_v1")
self.assertIsInstance(diagnostic["no_chord_reasons"], list)
def test_key_candidates_prioritize_expected_major_center(self):
chroma = normalize_vector(
np.array([0.34, 0.02, 0.06, 0.03, 0.16, 0.08, 0.03, 0.18, 0.02, 0.06, 0.01, 0.01], dtype=np.float32)
)
events = [
{"nome": "G", "inicio": 0.0, "fim": 1.0, "confianca": 0.9},
{"nome": "Em", "inicio": 1.0, "fim": 2.0, "confianca": 0.9},
{"nome": "C", "inicio": 2.0, "fim": 3.0, "confianca": 0.85},
{"nome": "D", "inicio": 3.0, "fim": 4.0, "confianca": 0.88},
{"nome": "G", "inicio": 4.0, "fim": 5.0, "confianca": 0.92},
]
root_histogram = normalize_vector(np.array([0.0, 0.0, 0.18, 0.0, 0.12, 0.0, 0.0, 0.46, 0.0, 0.0, 0.0, 0.24], dtype=np.float32))
candidates = build_key_candidates(chroma, root_histogram, events)
self.assertGreater(len(candidates), 0)
self.assertEqual(candidates[0].tonic, "G")
self.assertEqual(candidates[0].mode, "maior")
def test_rank_progression_respects_tonal_sequence(self):
segments = [
HarmonicSegment(index=0, start=0.0, end=1.0, duration=1.0, chroma=normalize_vector(np.array([0.3, 0, 0.02, 0, 0.2, 0, 0, 0.3, 0, 0.02, 0, 0.16], dtype=np.float32)), energy=0.3),
HarmonicSegment(index=1, start=1.0, end=2.0, duration=1.0, chroma=normalize_vector(np.array([0.18, 0, 0.02, 0.2, 0, 0, 0, 0.24, 0, 0.02, 0, 0.34], dtype=np.float32)), energy=0.34),
HarmonicSegment(index=2, start=2.0, end=3.0, duration=1.0, chroma=normalize_vector(np.array([0.28, 0, 0, 0, 0.02, 0.19, 0, 0.18, 0, 0, 0, 0.33], dtype=np.float32)), energy=0.33),
HarmonicSegment(index=3, start=3.0, end=4.0, duration=1.0, chroma=normalize_vector(np.array([0.02, 0, 0.28, 0, 0.02, 0.18, 0, 0.32, 0, 0, 0, 0.18], dtype=np.float32)), energy=0.32),
]
acoustic = [
[ChordCandidate("G", 7, "", 0.91, 0.92), ChordCandidate("C", 0, "", 0.89, 0.89)],
[ChordCandidate("Em", 4, "m", 1.38, 1.42), ChordCandidate("G", 7, "", 0.58, 0.6)],
[ChordCandidate("C", 0, "", 0.9, 0.91), ChordCandidate("Em", 4, "m", 0.83, 0.84)],
[ChordCandidate("D", 2, "", 1.05, 1.08), ChordCandidate("G", 7, "", 0.73, 0.76)],
]
ranked = rank_progression_for_key(segments, acoustic, KeyCandidate(tonic="G", mode="maior", confidence=0.82, score=5.0))
self.assertIsNotNone(ranked)
self.assertEqual([event["nome"] for event in ranked["events"]], ["G", "Em", "C", "D"])
def test_stage_progression_prefers_compact_playable_window(self):
events = merge_consecutive_events(
[
{"nome": "G", "inicio": 0.0, "fim": 1.0, "confianca": 0.92},
{"nome": "Em", "inicio": 1.0, "fim": 2.0, "confianca": 0.87},
{"nome": "C", "inicio": 2.0, "fim": 3.0, "confianca": 0.85},
{"nome": "D", "inicio": 3.0, "fim": 4.0, "confianca": 0.9},
{"nome": "G", "inicio": 4.0, "fim": 5.0, "confianca": 0.88},
],
min_duration=0.3,
)
stage = build_stage_progression(events, "G", "maior")
self.assertEqual(stage["progression"], "G Em C D")
self.assertEqual(stage["auxiliary"], "G Em C D G")
def test_estimate_beat_metrics_returns_stable_bpm(self):
beats = [0.0, 0.5, 1.0, 1.5, 2.0, 2.5]
metrics = estimate_beat_metrics(beats, duration=2.15)
self.assertAlmostEqual(metrics["bpm"], 120.0, places=1)
self.assertGreater(metrics["beat_confidence"], 0.5)
self.assertEqual(metrics["meter_hint"], "4/4")
def test_live_beat_eta_uses_last_beat_reference(self):
beats = [0.2, 0.7, 1.2, 1.7]
metrics = estimate_beat_metrics(beats, duration=2.2, live_mode=True)
self.assertGreater(metrics["next_beat_eta_ms"], 0)
self.assertLessEqual(metrics["next_beat_eta_ms"], metrics["beat_period_ms"])
def test_meter_hint_can_detect_three_four_pattern(self):
intervals = np.array([0.65, 0.45, 0.45, 0.65, 0.45, 0.45, 0.65, 0.45, 0.45], dtype=np.float32)
beat_times = np.concatenate([[0.0], np.cumsum(intervals)]).astype(np.float32)
meter = estimate_meter_hint(beat_times, 0.72)
self.assertEqual(meter, "3/4")
def test_candidate_payload_preserves_ambiguity_gap(self):
am = candidate_to_payload(ChordCandidate("Am", 9, "m", 1.02, 1.08))
dm = candidate_to_payload(ChordCandidate("Dm", 2, "m", 0.99, 1.01))
self.assertEqual(am["nome"], "Am")
self.assertLess(am["confianca"] - dm["confianca"], 0.05)
def test_bass_hint_resolves_am_vs_dm_candidate(self):
chroma = normalize_vector(
np.array([0.27, 0.0, 0.13, 0.0, 0.25, 0.03, 0.0, 0.0, 0.0, 0.32, 0.0, 0.0], dtype=np.float32)
)
candidates = build_segment_chord_candidates(
chroma,
top_k=3,
bass_hint=BassHint(pitch_pc=9, frequency_hz=110.0, confidence=0.82, energy=0.2),
)
self.assertGreaterEqual(len(candidates), 1)
self.assertEqual(candidates[0].name, "Am")
self.assertGreater(candidates[0].bass_score, 0.0)
def test_analisar_request_rejects_invalid_instrument(self):
with self.assertRaises(ValueError):
AnalisarRequest(path="/tmp/audio.wav", instrumento="guitarra")
if __name__ == "__main__":
unittest.main()
|