Add form metadata tags (#90)

* add form metadata tags

* format

aria/data/datasets.py

@@ -832,8 +832,6 @@ def _build(_midi_dataset):
 
  _build(_midi_dataset=midi_dataset)
 
- logger.info(
- f"Finished building, saved Finetuning to {save_path}"
- )
+ logger.info(f"Finished building, saved Finetuning to {save_path}")
 
  return cls(file_path=save_path, tokenizer=tokenizer)

aria/data/midi.py

@@ -14,6 +14,7 @@
 
 # TODO:
 # - Possibly refactor names 'mid' to 'midi'
+# - When pedal goes on after note on - this leads to it being played incorrectly.
 
 
 class MidiDict:
@@ -556,7 +557,7 @@ def get_duration_ms(
  return duration
 
 
-def _match_composer(text: str, composer_name: str):
+def _match_word(text: str, composer_name: str):
  # If name="bach" this pattern will match "bach", "Bach" or "BACH" if
  # it is either proceeded or preceded by a "_" or " ".
  pattern = (
@@ -581,7 +582,7 @@ def meta_composer_filename(
  file_name = pathlib.Path(mid.filename).stem
  matched_names = set()
  for name in composer_names:
- if _match_composer(file_name, name):
+ if _match_word(file_name, name):
  matched_names.add(name)
 
  # Only return data if only one composer is found
@@ -592,13 +593,28 @@ def meta_composer_filename(
  return {}
 
 
+def meta_form_filename(mid: mido.MidiFile, msg_data: dict, form_names: list):
+ file_name = pathlib.Path(mid.filename).stem
+ matched_names = set()
+ for name in form_names:
+ if _match_word(file_name, name):
+ matched_names.add(name)
+
+ # Only return data if only one composer is found
+ matched_names = list(matched_names)
+ if len(matched_names) == 1:
+ return {"form": matched_names[0]}
+ else:
+ return {}
+
+
 def meta_composer_metamsg(
  mid: mido.MidiFile, msg_data: dict, composer_names: list
 ):
  matched_names = set()
  for msg in msg_data["meta_msgs"]:
  for name in composer_names:
- if _match_composer(msg["data"], name):
+ if _match_word(msg["data"], name):
  matched_names.add(name)
 
  # Only return data if only one composer is found
@@ -614,6 +630,7 @@ def get_metadata_fn(metadata_proc_name: str):
  name_to_fn = {
  "composer_filename": meta_composer_filename,
  "composer_metamsg": meta_composer_metamsg,
+ "form_filename": meta_form_filename,
  }
 
  fn = name_to_fn.get(metadata_proc_name, None)

aria/sample.py

@@ -18,6 +18,7 @@
 
 # TODO: Add which instruments were detected in the prompt
 
+
 def _get_cfg_coeff(cfg_gamma, cfg_mode, cur_pos, start_pos, total_len):
  if cfg_mode is None:
  return cfg_gamma

aria/tokenizer/tokenizer.py

@@ -283,6 +283,8 @@ def __init__(self, return_tensors: bool = False):
  self.prefix_tokens += [
  ("prefix", "composer", x) for x in self.composer_names
  ]
+ self.form_names = self.config["form_names"]
+ self.prefix_tokens += [("prefix", "form", x) for x in self.form_names]
 
  # Build vocab
  self.time_tok = "<T>"
@@ -389,6 +391,10 @@ def tokenize_midi_dict(self, midi_dict: MidiDict):
  if composer and (composer in self.composer_names):
  prefix.insert(0, ("prefix", "composer", composer))
 
+ form = midi_dict.metadata.get("form")
+ if form and (form in self.form_names):
+ prefix.insert(0, ("prefix", "form", form))
+
  # NOTE: Any preceding silence is removed implicitly
  tokenized_seq = []
  initial_onset_tick = midi_dict.note_msgs[0]["data"]["start"]
@@ -919,6 +925,8 @@ def __init__(self, return_tensors: bool = False):
  self.prefix_tokens += [
  ("prefix", "composer", x) for x in self.composer_names
  ]
+ self.form_names = self.config["form_names"]
+ self.prefix_tokens += [("prefix", "form", x) for x in self.form_names]
 
  # Build vocab
  self.wait_tokens = [("wait", i) for i in self.time_step_quantizations]
@@ -1012,6 +1020,10 @@ def tokenize_midi_dict(self, midi_dict: MidiDict):
  if composer and (composer in self.composer_names):
  prefix.insert(0, ("prefix", "composer", composer))
 
+ form = midi_dict.metadata.get("form")
+ if form and (form in self.form_names):
+ prefix.insert(0, ("prefix", "form", form))
+
  # NOTE: Any preceding silence is removed implicitly
  tokenized_seq = []
  num_notes = len(midi_dict.note_msgs)

config/config.json

@@ -69,6 +69,12 @@
  "args": {
  "composer_names": ["bach", "beethoven", "mozart", "chopin", "rachmaninoff", "liszt", "debussy", "schubert", "brahms", "ravel", "satie", "scarlatti"]
  }
+ },
+ "form_filename": {
+ "run": true,
+ "args": {
+ "form_names": ["sonata", "prelude", "nocturne", "étude", "waltz", "mazurka", "impromptu", "fugue"]
+ }
  }
  }
  },
@@ -119,7 +125,8 @@
  "num_steps": 500,
  "step": 10
  },
- "composer_names": ["bach", "beethoven", "mozart", "chopin", "rachmaninoff", "liszt", "debussy", "schubert", "brahms", "ravel", "satie", "scarlatti"]
+ "composer_names": ["bach", "beethoven", "mozart", "chopin", "rachmaninoff", "liszt", "debussy", "schubert", "brahms", "ravel", "satie", "scarlatti"],
+ "form_names": ["sonata", "prelude", "nocturne", "étude", "waltz", "mazurka", "impromptu", "fugue"]
  },
  "abs": {
  "ignore_instruments": {
@@ -165,7 +172,8 @@
  "abs_time_step_ms": 5000,
  "max_dur_ms": 5000,
  "time_step_ms": 10,
- "composer_names": ["bach", "beethoven", "mozart", "chopin", "rachmaninoff", "liszt", "debussy", "schubert", "brahms", "ravel", "satie", "scarlatti"]
+ "composer_names": ["bach", "beethoven", "mozart", "chopin", "rachmaninoff", "liszt", "debussy", "schubert", "brahms", "ravel", "satie", "scarlatti"],
+ "form_names": ["sonata", "prelude", "nocturne", "étude", "waltz", "mazurka", "impromptu", "fugue"]
  }
  }
 }

tests/test_tokenizers.py

@@ -213,7 +213,7 @@ def tokenize_detokenize(file_name: str):
  tknzr = tokenizer.AbsTokenizer(return_tensors=False)
  tokenize_detokenize("basic.mid")
  tokenize_detokenize("arabesque.mid")
- tokenize_detokenize("beethoven.mid")
+ tokenize_detokenize("beethoven_sonata.mid")
  tokenize_detokenize("bach.mid")
  tokenize_detokenize("expressive.mid")
  tokenize_detokenize("pop.mid")
@@ -249,7 +249,7 @@ def tokenize_aug_detokenize(
  logging.info(f"pitch_aug_fn:\n{seq} ->\n\n{seq_pitch_augmented}\n")
  tokenize_aug_detokenize("basic.mid", pitch_aug_fn, "pitch")
  tokenize_aug_detokenize("arabesque.mid", pitch_aug_fn, "pitch")
- tokenize_aug_detokenize("beethoven.mid", pitch_aug_fn, "pitch")
+ tokenize_aug_detokenize("beethoven_sonata.mid", pitch_aug_fn, "pitch")
  tokenize_aug_detokenize("bach.mid", pitch_aug_fn, "pitch")
  tokenize_aug_detokenize("expressive.mid", pitch_aug_fn, "pitch")
  tokenize_aug_detokenize("pop.mid", pitch_aug_fn, "pitch")
@@ -264,7 +264,9 @@ def tokenize_aug_detokenize(
  )
  tokenize_aug_detokenize("basic.mid", velocity_aug_fn, "velocity")
  tokenize_aug_detokenize("arabesque.mid", velocity_aug_fn, "velocity")
- tokenize_aug_detokenize("beethoven.mid", velocity_aug_fn, "velocity")
+ tokenize_aug_detokenize(
+ "beethoven_sonata.mid", velocity_aug_fn, "velocity"
+ )
  tokenize_aug_detokenize("bach.mid", velocity_aug_fn, "velocity")
  tokenize_aug_detokenize("expressive.mid", velocity_aug_fn, "velocity")
  tokenize_aug_detokenize("pop.mid", velocity_aug_fn, "velocity")
@@ -283,7 +285,7 @@ def tokenize_aug_detokenize(
 
  tokenize_aug_detokenize("basic.mid", tempo_aug_fn, "tempo")
  tokenize_aug_detokenize("arabesque.mid", tempo_aug_fn, "tempo")
- tokenize_aug_detokenize("beethoven.mid", tempo_aug_fn, "tempo")
+ tokenize_aug_detokenize("beethoven_sonata.mid", tempo_aug_fn, "tempo")
  tokenize_aug_detokenize("bach.mid", tempo_aug_fn, "tempo")
  tokenize_aug_detokenize("expressive.mid", tempo_aug_fn, "tempo")
  tokenize_aug_detokenize("pop.mid", tempo_aug_fn, "tempo")
@@ -293,7 +295,7 @@ def tokenize_aug_detokenize(
 
  def test_aug_time(self):
  tknzr = tokenizer.AbsTokenizer()
- mid_dict = MidiDict.from_midi("tests/test_data/beethoven.mid")
+ mid_dict = MidiDict.from_midi("tests/test_data/beethoven_sonata.mid")
  tokenized_seq = tknzr.tokenize(mid_dict)[:4096]
  pitch_aug_fn = tknzr.export_pitch_aug(aug_range=5)
  velocity_aug_fn = tknzr.export_velocity_aug(aug_steps_range=2)
@@ -352,7 +354,7 @@ def tokenize_detokenize(file_name: str):
 
  tokenize_detokenize("basic.mid")
  tokenize_detokenize("arabesque.mid")
- tokenize_detokenize("beethoven.mid")
+ tokenize_detokenize("beethoven_sonata.mid")
  tokenize_detokenize("bach.mid")
  tokenize_detokenize("expressive.mid")
  tokenize_detokenize("pop.mid")
@@ -405,7 +407,7 @@ def test_aug(self):
 
  def test_aug_time(self):
  tknzr = tokenizer.RelTokenizer()
- mid_dict = MidiDict.from_midi("tests/test_data/beethoven.mid")
+ mid_dict = MidiDict.from_midi("tests/test_data/beethoven_sonata.mid")
  tokenized_seq = tknzr.tokenize(mid_dict)[:4096]
 
  pitch_aug_fn = tknzr.export_pitch_aug(aug_range=5)