Add Inference test using TTS API in all models unit tests

tests/tts_tests/test_align_tts_train.py

@@ -4,6 +4,7 @@ import shutil
 
 from tests import get_device_id, get_tests_output_path, run_cli
 from TTS.tts.configs.align_tts_config import AlignTTSConfig
+from TTS.utils.trainer_utils import get_last_checkpoint
 
 config_path = os.path.join(get_tests_output_path(), "test_model_config.json")
 output_path = os.path.join(get_tests_output_path(), "train_outputs")
@@ -47,6 +48,14 @@ run_cli(command_train)
 # Find latest folder
 continue_path = max(glob.glob(os.path.join(output_path, "*/")), key=os.path.getmtime)
 
+# Inference using TTS API
+continue_config_path = os.path.join(continue_path, "config.json")
+continue_restore_path, _ = get_last_checkpoint(continue_path)
+out_wav_path = os.path.join(get_tests_output_path(), 'output.wav')
+
+inference_command = f"CUDA_VISIBLE_DEVICES='{get_device_id()}' tts --text 'This is an example.' --config_path {continue_config_path} --model_path {continue_restore_path} --out_path {out_wav_path}"
+run_cli(inference_command)
+
 # restore the model and continue training for one more epoch
 command_train = f"CUDA_VISIBLE_DEVICES='{get_device_id()}' python TTS/bin/train_tts.py --continue_path {continue_path} "
 run_cli(command_train)

tests/tts_tests/test_fast_pitch_speaker_emb_train.py

@@ -5,6 +5,7 @@ import shutil
 from tests import get_device_id, get_tests_output_path, run_cli
 from TTS.config.shared_configs import BaseAudioConfig
 from TTS.tts.configs.fast_pitch_config import FastPitchConfig
+from TTS.utils.trainer_utils import get_last_checkpoint
 
 config_path = os.path.join(get_tests_output_path(), "test_fast_pitch_config.json")
 output_path = os.path.join(get_tests_output_path(), "train_outputs")
@@ -63,6 +64,16 @@ run_cli(command_train)
 # Find latest folder
 continue_path = max(glob.glob(os.path.join(output_path, "*/")), key=os.path.getmtime)
 
+# Inference using TTS API
+continue_config_path = os.path.join(continue_path, "config.json")
+continue_restore_path, _ = get_last_checkpoint(continue_path)
+out_wav_path = os.path.join(get_tests_output_path(), 'output.wav')
+speaker_id = "ljspeech-1"
+continue_speakers_path = os.path.join(continue_path, "speakers.json")
+
+inference_command = f"CUDA_VISIBLE_DEVICES='{get_device_id()}' tts --text 'This is an example.' --speaker_idx {speaker_id} --speakers_file_path {continue_speakers_path} --config_path {continue_config_path} --model_path {continue_restore_path} --out_path {out_wav_path}"
+run_cli(inference_command)
+
 # restore the model and continue training for one more epoch
 command_train = f"CUDA_VISIBLE_DEVICES='{get_device_id()}' python TTS/bin/train_tts.py --continue_path {continue_path} "
 run_cli(command_train)

tests/tts_tests/test_fast_pitch_train.py

@@ -5,6 +5,7 @@ import shutil
 from tests import get_device_id, get_tests_output_path, run_cli
 from TTS.config.shared_configs import BaseAudioConfig
 from TTS.tts.configs.fast_pitch_config import FastPitchConfig
+from TTS.utils.trainer_utils import get_last_checkpoint
 
 config_path = os.path.join(get_tests_output_path(), "test_fast_pitch_config.json")
 output_path = os.path.join(get_tests_output_path(), "train_outputs")
@@ -62,6 +63,14 @@ run_cli(command_train)
 # Find latest folder
 continue_path = max(glob.glob(os.path.join(output_path, "*/")), key=os.path.getmtime)
 
+# Inference using TTS API
+continue_config_path = os.path.join(continue_path, "config.json")
+continue_restore_path, _ = get_last_checkpoint(continue_path)
+out_wav_path = os.path.join(get_tests_output_path(), 'output.wav')
+
+inference_command = f"CUDA_VISIBLE_DEVICES='{get_device_id()}' tts --text 'This is an example.' --config_path {continue_config_path} --model_path {continue_restore_path} --out_path {out_wav_path}"
+run_cli(inference_command)
+
 # restore the model and continue training for one more epoch
 command_train = f"CUDA_VISIBLE_DEVICES='{get_device_id()}' python TTS/bin/train_tts.py --continue_path {continue_path} "
 run_cli(command_train)

tests/tts_tests/test_glow_tts_d-vectors_train.py

@@ -54,6 +54,16 @@ run_cli(command_train)
 # Find latest folder
 continue_path = max(glob.glob(os.path.join(output_path, "*/")), key=os.path.getmtime)
 
+# Inference using TTS API
+continue_config_path = os.path.join(continue_path, "config.json")
+continue_restore_path, _ = get_last_checkpoint(continue_path)
+out_wav_path = os.path.join(get_tests_output_path(), 'output.wav')
+speaker_id = "ljspeech-1"
+continue_speakers_path = config.d_vector_file
+
+inference_command = f"CUDA_VISIBLE_DEVICES='{get_device_id()}' tts --text 'This is an example.' --speaker_idx {speaker_id} --speakers_file_path {continue_speakers_path} --config_path {continue_config_path} --model_path {continue_restore_path} --out_path {out_wav_path}"
+run_cli(inference_command)
+
 # restore the model and continue training for one more epoch
 command_train = f"CUDA_VISIBLE_DEVICES='{get_device_id()}' python TTS/bin/train_tts.py --continue_path {continue_path} "
 run_cli(command_train)

tests/tts_tests/test_glow_tts_speaker_emb_train.py

@@ -51,6 +51,16 @@ run_cli(command_train)
 # Find latest folder
 continue_path = max(glob.glob(os.path.join(output_path, "*/")), key=os.path.getmtime)
 
+# Inference using TTS API
+continue_config_path = os.path.join(continue_path, "config.json")
+continue_restore_path, _ = get_last_checkpoint(continue_path)
+out_wav_path = os.path.join(get_tests_output_path(), 'output.wav')
+speaker_id = "ljspeech-1"
+continue_speakers_path = os.path.join(continue_path, "speakers.json")
+
+inference_command = f"CUDA_VISIBLE_DEVICES='{get_device_id()}' tts --text 'This is an example.' --speaker_idx {speaker_id} --speakers_file_path {continue_speakers_path} --config_path {continue_config_path} --model_path {continue_restore_path} --out_path {out_wav_path}"
+run_cli(inference_command)
+
 # restore the model and continue training for one more epoch
 command_train = f"CUDA_VISIBLE_DEVICES='{get_device_id()}' python TTS/bin/train_tts.py --continue_path {continue_path} "
 run_cli(command_train)

tests/tts_tests/test_glow_tts_train.py

@@ -4,6 +4,7 @@ import shutil
 
 from tests import get_device_id, get_tests_output_path, run_cli
 from TTS.tts.configs.glow_tts_config import GlowTTSConfig
+from TTS.utils.trainer_utils import get_last_checkpoint
 
 config_path = os.path.join(get_tests_output_path(), "test_model_config.json")
 output_path = os.path.join(get_tests_output_path(), "train_outputs")
@@ -49,6 +50,14 @@ run_cli(command_train)
 # Find latest folder
 continue_path = max(glob.glob(os.path.join(output_path, "*/")), key=os.path.getmtime)
 
+# Inference using TTS API
+continue_config_path = os.path.join(continue_path, "config.json")
+continue_restore_path, _ = get_last_checkpoint(continue_path)
+out_wav_path = os.path.join(get_tests_output_path(), 'output.wav')
+
+inference_command = f"CUDA_VISIBLE_DEVICES='{get_device_id()}' tts --text 'This is an example.' --config_path {continue_config_path} --model_path {continue_restore_path} --out_path {out_wav_path}"
+run_cli(inference_command)
+
 # restore the model and continue training for one more epoch
 command_train = f"CUDA_VISIBLE_DEVICES='{get_device_id()}' python TTS/bin/train_tts.py --continue_path {continue_path} "
 run_cli(command_train)

tests/tts_tests/test_speedy_speech_train.py

@@ -4,6 +4,7 @@ import shutil
 
 from tests import get_device_id, get_tests_output_path, run_cli
 from TTS.tts.configs.speedy_speech_config import SpeedySpeechConfig
+from TTS.utils.trainer_utils import get_last_checkpoint
 
 config_path = os.path.join(get_tests_output_path(), "test_speedy_speech_config.json")
 output_path = os.path.join(get_tests_output_path(), "train_outputs")
@@ -47,6 +48,14 @@ run_cli(command_train)
 # Find latest folder
 continue_path = max(glob.glob(os.path.join(output_path, "*/")), key=os.path.getmtime)
 
+# Inference using TTS API
+continue_config_path = os.path.join(continue_path, "config.json")
+continue_restore_path, _ = get_last_checkpoint(continue_path)
+out_wav_path = os.path.join(get_tests_output_path(), 'output.wav')
+
+inference_command = f"CUDA_VISIBLE_DEVICES='{get_device_id()}' tts --text 'This is an example for it.' --config_path {continue_config_path} --model_path {continue_restore_path} --out_path {out_wav_path}"
+run_cli(inference_command)
+
 # restore the model and continue training for one more epoch
 command_train = f"CUDA_VISIBLE_DEVICES='{get_device_id()}' python TTS/bin/train_tts.py --continue_path {continue_path} "
 run_cli(command_train)

tests/tts_tests/test_tacotron2_d-vectors_train.py

@@ -4,6 +4,7 @@ import shutil
 
 from tests import get_device_id, get_tests_output_path, run_cli
 from TTS.tts.configs.tacotron2_config import Tacotron2Config
+from TTS.utils.trainer_utils import get_last_checkpoint
 
 config_path = os.path.join(get_tests_output_path(), "test_model_config.json")
 output_path = os.path.join(get_tests_output_path(), "train_outputs")
@@ -52,6 +53,16 @@ run_cli(command_train)
 # Find latest folder
 continue_path = max(glob.glob(os.path.join(output_path, "*/")), key=os.path.getmtime)
 
+# Inference using TTS API
+continue_config_path = os.path.join(continue_path, "config.json")
+continue_restore_path, _ = get_last_checkpoint(continue_path)
+out_wav_path = os.path.join(get_tests_output_path(), 'output.wav')
+speaker_id = "ljspeech-1"
+continue_speakers_path = config.d_vector_file
+
+inference_command = f"CUDA_VISIBLE_DEVICES='{get_device_id()}' tts --text 'This is an example.' --speaker_idx {speaker_id} --speakers_file_path {continue_speakers_path} --config_path {continue_config_path} --model_path {continue_restore_path} --out_path {out_wav_path}"
+run_cli(inference_command)
+
 # restore the model and continue training for one more epoch
 command_train = f"CUDA_VISIBLE_DEVICES='{get_device_id()}' python TTS/bin/train_tts.py --continue_path {continue_path} "
 run_cli(command_train)

tests/tts_tests/test_tacotron2_speaker_emb_train.py

@@ -4,6 +4,7 @@ import shutil
 
 from tests import get_device_id, get_tests_output_path, run_cli
 from TTS.tts.configs.tacotron2_config import Tacotron2Config
+from TTS.utils.trainer_utils import get_last_checkpoint
 
 config_path = os.path.join(get_tests_output_path(), "test_model_config.json")
 output_path = os.path.join(get_tests_output_path(), "train_outputs")
@@ -49,6 +50,16 @@ run_cli(command_train)
 # Find latest folder
 continue_path = max(glob.glob(os.path.join(output_path, "*/")), key=os.path.getmtime)
 
+# Inference using TTS API
+continue_config_path = os.path.join(continue_path, "config.json")
+continue_restore_path, _ = get_last_checkpoint(continue_path)
+out_wav_path = os.path.join(get_tests_output_path(), 'output.wav')
+speaker_id = "ljspeech-1"
+continue_speakers_path = os.path.join(continue_path, "speakers.json")
+
+inference_command = f"CUDA_VISIBLE_DEVICES='{get_device_id()}' tts --text 'This is an example.' --speaker_idx {speaker_id} --speakers_file_path {continue_speakers_path} --config_path {continue_config_path} --model_path {continue_restore_path} --out_path {out_wav_path}"
+run_cli(inference_command)
+
 # restore the model and continue training for one more epoch
 command_train = f"CUDA_VISIBLE_DEVICES='{get_device_id()}' python TTS/bin/train_tts.py --continue_path {continue_path} "
 run_cli(command_train)

tests/tts_tests/test_tacotron2_train.py

@@ -4,6 +4,7 @@ import shutil
 
 from tests import get_device_id, get_tests_output_path, run_cli
 from TTS.tts.configs.tacotron2_config import Tacotron2Config
+from TTS.utils.trainer_utils import get_last_checkpoint
 
 config_path = os.path.join(get_tests_output_path(), "test_model_config.json")
 output_path = os.path.join(get_tests_output_path(), "train_outputs")
@@ -47,6 +48,14 @@ run_cli(command_train)
 # Find latest folder
 continue_path = max(glob.glob(os.path.join(output_path, "*/")), key=os.path.getmtime)
 
+# Inference using TTS API
+continue_config_path = os.path.join(continue_path, "config.json")
+continue_restore_path, _ = get_last_checkpoint(continue_path)
+out_wav_path = os.path.join(get_tests_output_path(), 'output.wav')
+
+inference_command = f"CUDA_VISIBLE_DEVICES='{get_device_id()}' tts --text 'This is an example.' --config_path {continue_config_path} --model_path {continue_restore_path} --out_path {out_wav_path}"
+run_cli(inference_command)
+
 # restore the model and continue training for one more epoch
 command_train = f"CUDA_VISIBLE_DEVICES='{get_device_id()}' python TTS/bin/train_tts.py --continue_path {continue_path} "
 run_cli(command_train)

tests/tts_tests/test_tacotron_train.py

@@ -4,6 +4,7 @@ import shutil
 
 from tests import get_device_id, get_tests_output_path, run_cli
 from TTS.tts.configs.tacotron_config import TacotronConfig
+from TTS.utils.trainer_utils import get_last_checkpoint
 
 config_path = os.path.join(get_tests_output_path(), "test_model_config.json")
 output_path = os.path.join(get_tests_output_path(), "train_outputs")
@@ -48,6 +49,14 @@ run_cli(command_train)
 # Find latest folder
 continue_path = max(glob.glob(os.path.join(output_path, "*/")), key=os.path.getmtime)
 
+# Inference using TTS API
+continue_config_path = os.path.join(continue_path, "config.json")
+continue_restore_path, _ = get_last_checkpoint(continue_path)
+out_wav_path = os.path.join(get_tests_output_path(), 'output.wav')
+
+inference_command = f"CUDA_VISIBLE_DEVICES='{get_device_id()}' tts --text 'This is an example.' --config_path {continue_config_path} --model_path {continue_restore_path} --out_path {out_wav_path}"
+run_cli(inference_command)
+
 # restore the model and continue training for one more epoch
 command_train = f"CUDA_VISIBLE_DEVICES='{get_device_id()}' python TTS/bin/train_tts.py --continue_path {continue_path} "
 run_cli(command_train)

tests/tts_tests/test_vits_multilingual_speaker_emb_train.py

@@ -5,6 +5,7 @@ import shutil
 from tests import get_device_id, get_tests_output_path, run_cli
 from TTS.config.shared_configs import BaseDatasetConfig
 from TTS.tts.configs.vits_config import VitsConfig
+from TTS.utils.trainer_utils import get_last_checkpoint
 
 config_path = os.path.join(get_tests_output_path(), "test_model_config.json")
 output_path = os.path.join(get_tests_output_path(), "train_outputs")
@@ -34,7 +35,7 @@ config = VitsConfig(
     text_cleaner="english_cleaners",
     use_phonemes=True,
     use_espeak_phonemes=True,
-    phoneme_language="en-us",
+    phoneme_language="en",
     phoneme_cache_path="tests/data/ljspeech/phoneme_cache/",
     run_eval=True,
     test_delay_epochs=-1,
@@ -82,6 +83,18 @@ run_cli(command_train)
 # Find latest folder
 continue_path = max(glob.glob(os.path.join(output_path, "*/")), key=os.path.getmtime)
 
+# Inference using TTS API
+continue_config_path = os.path.join(continue_path, "config.json")
+continue_restore_path, _ = get_last_checkpoint(continue_path)
+out_wav_path = os.path.join(get_tests_output_path(), 'output.wav')
+speaker_id = "ljspeech"
+languae_id = "en"
+continue_speakers_path = os.path.join(continue_path, "speakers.json")
+continue_languages_path = os.path.join(continue_path, "language_ids.json")
+
+inference_command = f"CUDA_VISIBLE_DEVICES='{get_device_id()}' tts --text 'This is an example.' --speaker_idx {speaker_id} --speakers_file_path {continue_speakers_path} --language_ids_file_path {continue_languages_path} --language_idx {languae_id} --config_path {continue_config_path} --model_path {continue_restore_path} --out_path {out_wav_path}"
+run_cli(inference_command)
+
 # restore the model and continue training for one more epoch
 command_train = f"CUDA_VISIBLE_DEVICES='{get_device_id()}' python TTS/bin/train_tts.py --continue_path {continue_path} "
 run_cli(command_train)

tests/tts_tests/test_vits_multilingual_train-d_vectors.py

@@ -5,13 +5,14 @@ import shutil
 from tests import get_device_id, get_tests_output_path, run_cli
 from TTS.config.shared_configs import BaseDatasetConfig
 from TTS.tts.configs.vits_config import VitsConfig
+from TTS.utils.trainer_utils import get_last_checkpoint
 
 config_path = os.path.join(get_tests_output_path(), "test_model_config.json")
 output_path = os.path.join(get_tests_output_path(), "train_outputs")
 
 
 dataset_config_en = BaseDatasetConfig(
-    name="ljspeech",
+    name="ljspeech_test",
     meta_file_train="metadata.csv",
     meta_file_val="metadata.csv",
     path="tests/data/ljspeech",
@@ -19,7 +20,7 @@ dataset_config_en = BaseDatasetConfig(
 )
 
 dataset_config_pt = BaseDatasetConfig(
-    name="ljspeech",
+    name="ljspeech_test",
     meta_file_train="metadata.csv",
     meta_file_val="metadata.csv",
     path="tests/data/ljspeech",
@@ -31,7 +32,7 @@ config = VitsConfig(
     eval_batch_size=2,
     num_loader_workers=0,
     num_eval_loader_workers=0,
-    text_cleaner="english_cleaners",
+    text_cleaner="multilingual_cleaners",
     use_phonemes=False,
     phoneme_cache_path="tests/data/ljspeech/phoneme_cache/",
     run_eval=True,
@@ -85,6 +86,18 @@ run_cli(command_train)
 # Find latest folder
 continue_path = max(glob.glob(os.path.join(output_path, "*/")), key=os.path.getmtime)
 
+# Inference using TTS API
+continue_config_path = os.path.join(continue_path, "config.json")
+continue_restore_path, _ = get_last_checkpoint(continue_path)
+out_wav_path = os.path.join(get_tests_output_path(), 'output.wav')
+speaker_id = "ljspeech-1"
+languae_id = "en"
+continue_speakers_path = config.d_vector_file
+continue_languages_path = os.path.join(continue_path, "language_ids.json")
+
+inference_command = f"CUDA_VISIBLE_DEVICES='{get_device_id()}' tts --text 'This is an example.' --speaker_idx {speaker_id} --speakers_file_path {continue_speakers_path} --language_ids_file_path {continue_languages_path} --language_idx {languae_id} --config_path {continue_config_path} --model_path {continue_restore_path} --out_path {out_wav_path}"
+run_cli(inference_command)
+
 # restore the model and continue training for one more epoch
 command_train = f"CUDA_VISIBLE_DEVICES='{get_device_id()}' python TTS/bin/train_tts.py --continue_path {continue_path} "
 run_cli(command_train)

tests/tts_tests/test_vits_speaker_emb_train.py

@@ -4,6 +4,7 @@ import shutil
 
 from tests import get_device_id, get_tests_output_path, run_cli
 from TTS.tts.configs.vits_config import VitsConfig
+from TTS.utils.trainer_utils import get_last_checkpoint
 
 config_path = os.path.join(get_tests_output_path(), "test_model_config.json")
 output_path = os.path.join(get_tests_output_path(), "train_outputs")
@@ -25,7 +26,7 @@ config = VitsConfig(
     print_step=1,
     print_eval=True,
     test_sentences=[
-        ["Be a voice, not an echo.", "ljspeech"],
+        ["Be a voice, not an echo.", "ljspeech-1"],
     ],
 )
 # set audio config
@@ -45,7 +46,7 @@ config.save_json(config_path)
 command_train = (
     f"CUDA_VISIBLE_DEVICES='{get_device_id()}' python TTS/bin/train_tts.py --config_path {config_path} "
     f"--coqpit.output_path {output_path} "
-    "--coqpit.datasets.0.name ljspeech "
+    "--coqpit.datasets.0.name ljspeech_test "
     "--coqpit.datasets.0.meta_file_train metadata.csv "
     "--coqpit.datasets.0.meta_file_val metadata.csv "
     "--coqpit.datasets.0.path tests/data/ljspeech "
@@ -57,6 +58,16 @@ run_cli(command_train)
 # Find latest folder
 continue_path = max(glob.glob(os.path.join(output_path, "*/")), key=os.path.getmtime)
 
+# Inference using TTS API
+continue_config_path = os.path.join(continue_path, "config.json")
+continue_restore_path, _ = get_last_checkpoint(continue_path)
+out_wav_path = os.path.join(get_tests_output_path(), 'output.wav')
+speaker_id = "ljspeech-1"
+continue_speakers_path = os.path.join(continue_path, "speakers.json")
+
+inference_command = f"CUDA_VISIBLE_DEVICES='{get_device_id()}' tts --text 'This is an example.' --speaker_idx {speaker_id} --speakers_file_path {continue_speakers_path} --config_path {continue_config_path} --model_path {continue_restore_path} --out_path {out_wav_path}"
+run_cli(inference_command)
+
 # restore the model and continue training for one more epoch
 command_train = f"CUDA_VISIBLE_DEVICES='{get_device_id()}' python TTS/bin/train_tts.py --continue_path {continue_path} "
 run_cli(command_train)

tests/tts_tests/test_vits_train.py

@@ -4,6 +4,7 @@ import shutil
 
 from tests import get_device_id, get_tests_output_path, run_cli
 from TTS.tts.configs.vits_config import VitsConfig
+from TTS.utils.trainer_utils import get_last_checkpoint
 
 config_path = os.path.join(get_tests_output_path(), "test_model_config.json")
 output_path = os.path.join(get_tests_output_path(), "train_outputs")
@@ -48,6 +49,14 @@ run_cli(command_train)
 # Find latest folder
 continue_path = max(glob.glob(os.path.join(output_path, "*/")), key=os.path.getmtime)
 
+# Inference using TTS API
+continue_config_path = os.path.join(continue_path, "config.json")
+continue_restore_path, _ = get_last_checkpoint(continue_path)
+out_wav_path = os.path.join(get_tests_output_path(), 'output.wav')
+
+inference_command = f"CUDA_VISIBLE_DEVICES='{get_device_id()}' tts --text 'This is an example.' --config_path {continue_config_path} --model_path {continue_restore_path} --out_path {out_wav_path}"
+run_cli(inference_command)
+
 # restore the model and continue training for one more epoch
 command_train = f"CUDA_VISIBLE_DEVICES='{get_device_id()}' python TTS/bin/train_tts.py --continue_path {continue_path} "
 run_cli(command_train)