def match_one_midi(midi_file):
'''
Hash and match a single MIDI file against the MSD
:parameters:
- midi_file : str
Path to a MIDI file to match
'''
# Get a beat-synchronous piano roll of the MIDI
pm = pretty_midi.PrettyMIDI(midi_file)
piano_roll = pm.get_piano_roll(times=pm.get_beats()).T
piano_roll = piano_roll[np.newaxis, :, 36:84]
# Make the piano roll look like it does when we trained the hasher
piano_roll = (piano_roll - train_stats['mean'])/train_stats['std']
hashed_piano_roll = hash(
piano_roll[np.newaxis].astype(theano.config.floatX))
# Compute hash sequence
query = hash_match.vectors_to_ints(hashed_piano_roll > 0)
query = query.astype('uint16')
# Get indices of sequences which are within 40% -> 1/40% of this seq length
valid_length_indices = hash_match.filter_by_length(query, sequences, .4)
# Compute MIDI mean chroma vector
query_chroma = pm.get_chroma().mean(axis=1)
# Get sequences less than the mean chroma distance
valid_chroma_indices = hash_match.filter_by_mean_chroma(
query_chroma, mean_chromas, 20)
# Intersect to get valid index set
valid_indices = np.intersect1d(valid_length_indices, valid_chroma_indices)
# Match the MIDI file query hash list against all sequences
matches, scores = hash_match.match_one_sequence(
query, sequences, .9, 4, valid_indices)
return matches, scores
def match_one_midi(midi_data, msd_match_indices):
'''
Match a MIDI sequence against the MSD and evaluate whether a good match was
found
:parameters:
- midi_data : dict
Dict of MIDI data, including mean chroma vector, hash sequence, etc
- msd_match_indices : list-like of int
indices of entries in the MSD this MIDI should potentially match to
:returns:
- results : dict
Dictionary with diagnostics about whether this match was successful
'''
# Match this MIDI sequence against MSD sequences
matches, scores, n_pruned_dist = hash_match.match_one_sequence(
midi_data['hash_list'], sequences, GULLY, PENALTY)
# Store results of the match
results = {}
results['midi_md5'] = midi_data['md5']
results['msd_match_ids'] = [data[n]['id'] for n in msd_match_indices]
# Compile the rank and score for each MSD entry which should match the MIDI
matched_ranks = []
matched_scores = []
for msd_index in msd_match_indices:
# If it was pruned by mean chroma/length pruning, compute manually
if msd_index not in matches:
_, score, _ = hash_match.match_one_sequence(
midi_data['hash_list'], [sequences[msd_index]], GULLY, PENALTY)
score = score[0]
rank = (scores < score).sum()
# Otherwise find the result
else:
rank = matches.index(msd_index)
score = scores[rank]
matched_ranks.append(rank)
matched_scores.append(score)
results['msd_match_ranks'] = matched_ranks
results['msd_match_scores'] = matched_scores
results['n_pruned_dist'] = n_pruned_dist
return results
