[TTS][refactor] Part 3 - nemo.collections.tts.g2p.models

Jenkinsfile

@@ -1035,7 +1035,7 @@ pipeline {
               steps {
                 sh 'cd examples/tts/g2p && \
                     TIME=`date +"%Y-%m-%d-%T"` && OUTPUT_DIR=output_${TIME} && \
-                    python heteronym_classification_train_and_evaluate.py \
+                    python g2p_heteronym_classification_train_and_evaluate.py \
                         train_manifest=/home/TestData/g2p/manifest.json \
                         validation_manifest=/home/TestData/g2p/manifest.json \
                         test_manifest=/home/TestData/g2p/manifest.json \
@@ -1047,7 +1047,7 @@ pipeline {
                         exp_manager.exp_dir=${OUTPUT_DIR} \
                         +exp_manager.use_datetime_version=False\
                         +exp_manager.version=test && \
-                    python heteronym_classification_inference.py \
+                    python g2p_heteronym_classification_inference.py \
                         manifest=/home/TestData/g2p/manifest.json \
                         pretrained_model=${OUTPUT_DIR}/HeteronymClassification/test/checkpoints/HeteronymClassification.nemo \
                         output_manifest=preds.json'

docs/source/tts/g2p.rst

@@ -59,7 +59,7 @@ To train ByT5 G2P model and evaluate it after at the end of the training, run:
         do_training=True \
         do_testing=True
 
-Example of the config file: ``NeMo/examples/text_processing/g2p/conf/t5_g2p.yaml``.
+Example of the config file: ``NeMo/examples/tts/g2p/conf/g2p_t5.yaml``.
 
 
 To train G2P-Conformer model and evaluate it after at the end of the training, run:
@@ -168,7 +168,7 @@ To train the model, run:
 
 .. code-block::
 
-    python heteronym_classification_train_and_evaluate.py \
+    python g2p_heteronym_classification_train_and_evaluate.py \
         train_manifest=<Path to train manifest file>" \
         validation_manifest=<Path to validation manifest file>" \
         model.wordids=<Path to wordids.tsv file, similar to https://github.com/google-research-datasets/WikipediaHomographData/blob/master/data/wordids.tsv> \
@@ -179,7 +179,7 @@ To train the model and evaluate it when the training is complete, run:
 
 .. code-block::
 
-    python heteronym_classification_train_and_evaluate.py \
+    python g2p_heteronym_classification_train_and_evaluate.py \
         train_manifest=<Path to train manifest file>" \
         validation_manifest=<Path to validation manifest file>" \
         model.test_ds.dataset.manifest=<Path to test manifest file>" \
@@ -191,7 +191,7 @@ To evaluate pretrained model, run:
 
 .. code-block::
 
-    python heteronym_classification_train_and_evaluate.py \
+    python g2p_heteronym_classification_train_and_evaluate.py \
         do_training=False \
         do_testing=True \
         model.test_ds.dataset.manifest=<Path to test manifest file>"  \
@@ -201,7 +201,7 @@ To run inference with a pretrained HeteronymClassificationModel, run:
 
 .. code-block::
 
-    python heteronym_classification_inference.py \
+    python g2p_heteronym_classification_inference.py \
         manifest="<Path to .json manifest>" \
         pretrained_model="<Path to .nemo file or pretrained model name from list_available_models()>" \
         output_file="<Path to .json manifest to save prediction>"

examples/tts/g2p/heteronym_classification_inference.py → examples/tts/g2p/g2p_heteronym_classification_inference.py

@@ -22,7 +22,7 @@
 import torch
 from omegaconf import OmegaConf
 
-from nemo.collections.tts.g2p.models.heteronym_classification import HeteronymClassificationModel
+from nemo.collections.tts.models.g2p_heteronym_classification import HeteronymClassificationModel
 from nemo.core.config import hydra_runner
 from nemo.utils import logging
 
@@ -34,15 +34,15 @@
 
 Inference form manifest:
 
-python heteronym_classification_inference.py \
+python g2p_heteronym_classification_inference.py \
     manifest="<Path to .json manifest>" \
     pretrained_model="<Path to .nemo file or pretrained model name from list_available_models()>" \
     output_manifest="<Path to .json manifest to save prediction>" \
     wordid_to_phonemes_file="<Path to a file with mapping from wordid predicted by the model to phonemes>"
 
 Interactive inference:
 
-python heteronym_classification_inference.py \
+python g2p_heteronym_classification_inference.py \
     pretrained_model="<Path to .nemo file or pretrained model name from list_available_models()>" \
     wordid_to_phonemes_file="<Path to a file with mapping from wordid predicted by the model to phonemes>" # Optional
 

examples/tts/g2p/heteronym_classification_train_and_evaluate.py → examples/tts/g2p/g2p_heteronym_classification_train_and_evaluate.py

@@ -18,7 +18,7 @@
 import torch
 
 from nemo.collections.common.callbacks import LogEpochTimeCallback
-from nemo.collections.tts.g2p.models.heteronym_classification import HeteronymClassificationModel
+from nemo.collections.tts.models.g2p_heteronym_classification import HeteronymClassificationModel
 from nemo.core.config import hydra_runner
 from nemo.utils import logging
 from nemo.utils.exp_manager import exp_manager
@@ -29,14 +29,14 @@
 To prepare dataset, see NeMo/scripts/dataset_processing/g2p/export_wikihomograph_data_to_manifest.py
 
 To run training:
-python heteronym_classification_train_and_evaluate.py \
+python g2p_heteronym_classification_train_and_evaluate.py \
     train_manifest=<Path to train manifest file>" \
     validation_manifest=<Path to validation manifest file>" \
     model.wordids="<Path to wordids.tsv file>" \
     do_training=True
 
 To run training and testing (once the training is complete):
-python heteronym_classification_train_and_evaluate.py \
+python g2p_heteronym_classification_train_and_evaluate.py \
     train_manifest=<Path to train manifest file>" \
     validation_manifest=<Path to validation manifest file>" \
     model.test_ds.dataset.manifest=<Path to test manifest file>" \
@@ -45,7 +45,7 @@
     do_testing=True
 
 To run testing:
-python heteronym_classification_train_and_evaluate.py \
+python g2p_heteronym_classification_train_and_evaluate.py \
     do_training=False \
     do_testing=True \
     model.test_ds.dataset.manifest=<Path to test manifest file>"  \
@@ -60,7 +60,7 @@
 """
 
 
-@hydra_runner(config_path="conf", config_name="heteronym_classification.yaml")
+@hydra_runner(config_path="conf", config_name="g2p_heteronym_classification.yaml")
 def main(cfg):
     trainer = pl.Trainer(**cfg.trainer)
     exp_manager(trainer, cfg.get("exp_manager", None))

examples/tts/g2p/g2p_inference.py

@@ -21,7 +21,7 @@
 from omegaconf import OmegaConf
 from utils import get_metrics
 
-from nemo.collections.tts.g2p.models.g2p_model import G2PModel
+from nemo.collections.tts.models.base import G2PModel
 from nemo.core.config import hydra_runner
 from nemo.utils import logging
 

examples/tts/g2p/g2p_train_and_evaluate.py

@@ -19,14 +19,14 @@
 from utils import get_model
 
 from nemo.collections.common.callbacks import LogEpochTimeCallback
-from nemo.collections.tts.g2p.models.g2p_model import G2PModel
+from nemo.collections.tts.models.base import G2PModel
 from nemo.core.config import hydra_runner
 from nemo.utils import logging, model_utils
 from nemo.utils.exp_manager import exp_manager
 
 """
 This script supports training of G2PModels 
-(for T5G2PModel use t5_g2p.yaml, for CTCG2PModel use either g2p_conformer.yaml or g2p_t5_ctc.yaml)
+(for T5G2PModel use g2p_t5.yaml, for CTCG2PModel use either g2p_conformer.yaml or g2p_t5_ctc.yaml)
 
 # Training T5G2PModel and evaluation at the end of training:
     python examples/text_processing/g2p/g2p_train_and_evaluate.py \
@@ -38,7 +38,7 @@
         do_training=True \
         do_testing=True
 
-    Example of the config file: NeMo/examples/text_processing/g2p/conf/t5_g2p.yaml
+    Example of the config file: NeMo/examples/tts/g2p/conf/g2p_t5.yaml
 
 # Training Conformer-G2P Model and evaluation at the end of training:
     python examples/text_processing/g2p/g2p_train_and_evaluate.py \
@@ -64,7 +64,7 @@
 """
 
 
-@hydra_runner(config_path="conf", config_name="t5_g2p")
+@hydra_runner(config_path="conf", config_name="g2p_t5")
 def main(cfg):
     trainer = pl.Trainer(**cfg.trainer)
     exp_manager(trainer, cfg.get("exp_manager", None))

examples/tts/g2p/utils.py

@@ -15,8 +15,8 @@
 import json
 
 from nemo.collections.asr.metrics.wer import word_error_rate
-from nemo.collections.tts.g2p.models import CTCG2PModel
-from nemo.collections.tts.g2p.models.t5_g2p import T5G2PModel
+from nemo.collections.tts.models.g2p_ctc import CTCG2PModel
+from nemo.collections.tts.models.g2p_t5 import T5G2PModel
 from nemo.utils import logging
 
 

nemo/collections/tts/g2p/modules.py

@@ -79,7 +79,7 @@ def setup_heteronym_model(
         """
 
         try:
-            from nemo.collections.tts.g2p.models.heteronym_classification import HeteronymClassificationModel
+            from nemo.collections.tts.models.g2p_heteronym_classification import HeteronymClassificationModel
 
             self.heteronym_model = heteronym_model
             self.heteronym_model.set_wordid_to_phonemes(wordid_to_phonemes_file)