fix(configs): update config field types

TTS/tts/configs/align_tts_config.py

@@ -69,7 +69,7 @@ class AlignTTSConfig(BaseTTSConfig):
     model: str = "align_tts"
     # model specific params
     model_args: AlignTTSArgs = field(default_factory=AlignTTSArgs)
-    phase_start_steps: list[int] = None
+    phase_start_steps: list[int] | None = None
 
     ssim_alpha: float = 1.0
     spec_loss_alpha: float = 1.0
@@ -79,13 +79,13 @@ class AlignTTSConfig(BaseTTSConfig):
     # multi-speaker settings
     use_speaker_embedding: bool = False
     use_d_vector_file: bool = False
-    d_vector_file: str = False
+    d_vector_file: str | None = None
 
     # optimizer parameters
     optimizer: str = "Adam"
     optimizer_params: dict = field(default_factory=lambda: {"betas": [0.9, 0.998], "weight_decay": 1e-6})
-    lr_scheduler: str = None
-    lr_scheduler_params: dict = None
+    lr_scheduler: str | None = None
+    lr_scheduler_params: dict | None = None
     lr: float = 1e-4
     grad_clip: float = 5.0
 

TTS/tts/configs/fast_pitch_config.py

@@ -116,10 +116,10 @@ class FastPitchConfig(BaseTTSConfig):
 
     # multi-speaker settings
     num_speakers: int = 0
-    speakers_file: str = None
+    speakers_file: str | None = None
     use_speaker_embedding: bool = False
     use_d_vector_file: bool = False
-    d_vector_file: str = False
+    d_vector_file: str | None = None
     d_vector_dim: int = 0
 
     # optimizer parameters
@@ -149,7 +149,7 @@ class FastPitchConfig(BaseTTSConfig):
 
     # dataset configs
     compute_f0: bool = True
-    f0_cache_path: str = None
+    f0_cache_path: str | None = None
 
     # testing
     test_sentences: list[str] = field(

TTS/tts/configs/fast_speech_config.py

@@ -110,10 +110,10 @@ class FastSpeechConfig(BaseTTSConfig):
 
     # multi-speaker settings
     num_speakers: int = 0
-    speakers_file: str = None
+    speakers_file: str | None = None
     use_speaker_embedding: bool = False
     use_d_vector_file: bool = False
-    d_vector_file: str = False
+    d_vector_file: str | None = None
     d_vector_dim: int = 0
 
     # optimizer parameters
@@ -143,7 +143,7 @@ class FastSpeechConfig(BaseTTSConfig):
 
     # dataset configs
     compute_f0: bool = False
-    f0_cache_path: str = None
+    f0_cache_path: str | None = None
 
     # testing
     test_sentences: list[str] = field(

TTS/tts/configs/fastspeech2_config.py

@@ -126,10 +126,10 @@ class Fastspeech2Config(BaseTTSConfig):
 
     # multi-speaker settings
     num_speakers: int = 0
-    speakers_file: str = None
+    speakers_file: str | None = None
     use_speaker_embedding: bool = False
     use_d_vector_file: bool = False
-    d_vector_file: str = False
+    d_vector_file: str | None = None
     d_vector_dim: int = 0
 
     # optimizer parameters
@@ -160,11 +160,11 @@ class Fastspeech2Config(BaseTTSConfig):
 
     # dataset configs
     compute_f0: bool = True
-    f0_cache_path: str = None
+    f0_cache_path: str | None = None
 
     # dataset configs
     compute_energy: bool = True
-    energy_cache_path: str = None
+    energy_cache_path: str | None = None
 
     # testing
     test_sentences: list[str] = field(

TTS/tts/configs/glow_tts_config.py

@@ -100,7 +100,7 @@ class GlowTTSConfig(BaseTTSConfig):
     model: str = "glow_tts"
 
     # model params
-    num_chars: int = None
+    num_chars: int | None = None
     encoder_type: str = "rel_pos_transformer"
     encoder_params: dict = field(
         default_factory=lambda: {
@@ -146,15 +146,15 @@ class GlowTTSConfig(BaseTTSConfig):
     data_dep_init_steps: int = 10
 
     # inference params
-    style_wav_for_test: str = None
+    style_wav_for_test: str | None = None
     inference_noise_scale: float = 0.0
     length_scale: float = 1.0
 
     # multi-speaker settings
     use_speaker_embedding: bool = False
-    speakers_file: str = None
+    speakers_file: str | None = None
     use_d_vector_file: bool = False
-    d_vector_file: str = False
+    d_vector_file: str | None = None
 
     # optimizer parameters
     optimizer: str = "RAdam"

TTS/tts/configs/speedy_speech_config.py

@@ -128,10 +128,10 @@ class SpeedySpeechConfig(BaseTTSConfig):
 
     # multi-speaker settings
     num_speakers: int = 0
-    speakers_file: str = None
+    speakers_file: str | None = None
     use_speaker_embedding: bool = False
     use_d_vector_file: bool = False
-    d_vector_file: str = False
+    d_vector_file: str | None = None
     d_vector_dim: int = 0
 
     # optimizer parameters
@@ -160,7 +160,7 @@ class SpeedySpeechConfig(BaseTTSConfig):
 
     # dataset configs
     compute_f0: bool = False
-    f0_cache_path: str = None
+    f0_cache_path: str | None = None
 
     # testing
     test_sentences: list[str] = field(

TTS/tts/configs/tacotron_config.py

@@ -169,7 +169,7 @@ class TacotronConfig(BaseTTSConfig):
 
     # attention layers
     attention_type: str = "original"
-    attention_heads: int = None
+    attention_heads: int | None = None
     attention_norm: str = "sigmoid"
     attention_win: bool = False
     windowing: bool = False
@@ -188,8 +188,8 @@ class TacotronConfig(BaseTTSConfig):
     use_speaker_embedding: bool = False
     speaker_embedding_dim: int = 512
     use_d_vector_file: bool = False
-    d_vector_file: str = False
-    d_vector_dim: int = None
+    d_vector_file: str | None = None
+    d_vector_dim: int | None = None
 
     # optimizer parameters
     optimizer: str = "RAdam"

TTS/tts/configs/vits_config.py

@@ -145,7 +145,7 @@ class VitsConfig(BaseTTSConfig):
     add_blank: bool = True
 
     # testing
-    test_sentences: list[list] = field(
+    test_sentences: list[str] | list[list[str]] = field(
         default_factory=lambda: [
             ["It took me quite a long time to develop a voice, and now that I have it I'm not going to be silent."],
             ["Be a voice, not an echo."],

TTS/vocoder/configs/multiband_melgan_config.py

@@ -121,7 +121,7 @@ class MultibandMelganConfig(BaseGANVocoderConfig):
     pad_short: int = 2000
     use_noise_augment: bool = False
     use_cache: bool = True
-    steps_to_start_discriminator: bool = 200000
+    steps_to_start_discriminator: int = 200000
 
     # LOSS PARAMETERS - overrides
     use_stft_loss: bool = True

TTS/vocoder/configs/shared_configs.py

@@ -168,7 +168,7 @@ class BaseGANVocoderConfig(BaseVocoderConfig):
     target_loss: str = "loss_0"  # loss value to pick the best model to save after each epoch
 
     # optimizer
-    grad_clip: float = field(default_factory=lambda: [5, 5])
+    grad_clip: float | list[float] = field(default_factory=lambda: [5, 5])
     lr_gen: float = 0.0002  # Initial learning rate.
     lr_disc: float = 0.0002  # Initial learning rate.
     lr_scheduler_gen: str = "ExponentialLR"  # one of the schedulers from https:#pytorch.org/docs/stable/optim.html