Update for japanese model (#706)

MontrealCorpusTools · Oct 12, 2023 · 4530f97 · 4530f97
1 parent 4730e9b
commit 4530f97
Show file tree

Hide file tree

Showing 44 changed files with 3,105 additions and 1,020 deletions.
diff --git a/docs/source/changelog/changelog_3.0.rst b/docs/source/changelog/changelog_3.0.rst
@@ -5,23 +5,31 @@
 3.0 Changelog
 *************
 
+3.0.0a6
+=======
+
+- Added support for generating pronunciations during training and alignment via :code:`--g2p_model_path`
+- Added support for Japanese tokenization through sudachipy
+- Fixed a crash in fine tuning
+- Added functionality for allowing a directory to be passed as the output path for :ref:`align_one`
+
 3.0.0a5
 =======
 
-- Update for :xref:`kalpy` version 0.5.5
-- Update :code:`--single_speaker` mode to not perform speaker adaptation
-- Add documentation for :ref:`concept_speaker_adaptation`
+- Updated for :xref:`kalpy` version 0.5.5
+- Updated :code:`--single_speaker` mode to not perform speaker adaptation
+- Added documentation for :ref:`concept_speaker_adaptation`
 
 3.0.0a4
 =======
 
-- Separate out segmentation functionality into :ref:`create_segments` and :ref:`create_segments_vad`
-- Fix a bug in :ref:`align_one` when specifying a :code:`config_path`
+- Separated out segmentation functionality into :ref:`create_segments` and :ref:`create_segments_vad`
+- Fixed a bug in :ref:`align_one` when specifying a :code:`config_path`
 
 3.0.0a3
 =======
 
-- Refactor tokenization for future spacy use
+- Refactored tokenization for future spacy use
 
 3.0.0a2
 =======

diff --git a/montreal_forced_aligner/abc.py b/montreal_forced_aligner/abc.py
@@ -392,7 +392,7 @@ def construct_engine(self, **kwargs) -> sqlalchemy.engine.Engine:
         return e
 
     @property
-    def session(self, **kwargs) -> sqlalchemy.orm.scoped_session:
+    def session(self) -> sqlalchemy.orm.scoped_session:
         """
         Construct database session
 
@@ -408,7 +408,7 @@ def session(self, **kwargs) -> sqlalchemy.orm.scoped_session:
         """
         if self._session is None:
             self._session = scoped_session(
-                sessionmaker(bind=self.db_engine, expire_on_commit=False, **kwargs)
+                sessionmaker(bind=self.db_engine, expire_on_commit=False)
             )
         return self._session
 

diff --git a/montreal_forced_aligner/acoustic_modeling/base.py b/montreal_forced_aligner/acoustic_modeling/base.py
@@ -488,6 +488,9 @@ def meta(self) -> MetaDict:
             utterance_count, duration, average_log_likelihood = summary.first()
         data = {
             "phones": sorted(self._generate_non_positional_list(self.non_silence_phones)),
+            "phone_mapping": {
+                k: v for k, v in self.phone_mapping.items() if not k.startswith("#")
+            },
             "phone_groups": self.worker.phone_groups,
             "version": get_mfa_version(),
             "architecture": self.architecture,
@@ -510,6 +513,7 @@ def meta(self) -> MetaDict:
                 "clitic_marker": self.worker.clitic_marker,
                 "position_dependent_phones": self.worker.position_dependent_phones,
             },
+            "language": str(self.worker.language),
             "features": self.feature_options,
             "oov_phone": self.worker.oov_phone,
             "optional_silence_phone": self.worker.optional_silence_phone,

diff --git a/montreal_forced_aligner/acoustic_modeling/trainer.py b/montreal_forced_aligner/acoustic_modeling/trainer.py
@@ -441,9 +441,9 @@ def export_model(self, output_model_path: Path) -> None:
             export_directory = os.path.dirname(output_model_path)
             if export_directory:
                 os.makedirs(export_directory, exist_ok=True)
-            # self.export_trained_rules(
-            #    self.training_configs[self.final_identifier].working_directory
-            # )
+            self.export_trained_rules(
+                self.training_configs[self.final_identifier].working_directory
+            )
             with self.session() as session:
                 for d in session.query(Dictionary):
                     base_name = self.dictionary_base_names[d.id]

diff --git a/montreal_forced_aligner/acoustic_modeling/triphone.py b/montreal_forced_aligner/acoustic_modeling/triphone.py
@@ -392,12 +392,17 @@ def _setup_tree(self, init_from_previous=False, initial_mix_up=True) -> None:
                 if initial_mix_up:
                     mix_up = self.initial_gaussians
                     mix_down = self.initial_gaussians
+                old_transition_model, old_acoustic_model = read_gmm_model(
+                    os.path.join(self.previous_aligner.working_directory, "final.mdl")
+                )
+                old_tree = read_tree(os.path.join(self.previous_aligner.working_directory, "tree"))
                 gmm_init_model_from_previous(
                     topo,
                     tree,
                     tree_stats,
-                    os.path.join(self.previous_aligner.working_directory, "final.mdl"),
-                    os.path.join(self.previous_aligner.working_directory, "tree"),
+                    old_acoustic_model,
+                    old_transition_model,
+                    old_tree,
                     str(self.model_path),
                     mixup=mix_up,
                     mixdown=mix_down,